unet.md

• 640x320 / misc
• 256
  • evaluation @ 256
    • vis @ evaluation/256
    • hml @ evaluation/256
  • vis @ 256 - hml @ vis/256
  • validation @ 256
  • videos @ 256
  • new @ 256
• 384
  • evaluation @ 384
    • vis @ evaluation/384
    • hml @ evaluation/384
  • validation @ 384
  • videos @ 384
• 512
  • evaluation @ 512
    • hml @ evaluation/512
  • vis @ 512 - hml @ vis/512
  • validation @ 512 - stitching @ validation/512
  • videos @ 512
• 640
  • 4__non_aug @ 640
  • 4 @ 640
    • evaluation @ 4/640
    • 4_49 @ 4/640
    • no_aug @ 4/640
      • stitched @ no_aug/4/640
    • no_aug__4_49 @ 4/640
  • 8 @ 640
    • evaluation @ 8/640
    • no_aug @ 8/640
      • stitched @ no_aug/8/640
  • 16 @ 640
    • continue_133 @ 16/640
    • continue_latest @ 16/640
    • evaluation @ 16/640
    • no_aug @ 16/640
      • stitched @ no_aug/16/640
  • 24 @ 640
    • continue_171 @ 24/640
    • continue_latest @ 24/640
    • evaluation @ 24/640
    • no_aug @ 24/640
      • stitched @ no_aug/24/640
  • 32 @ 640
    • evaluation @ 32/640
    • vis @ 32/640
    • no_aug @ 32/640
      • stitched @ no_aug/32/640
    • validation @ 32/640
    • stitching @ 32/640
    • YUN00001 @ 32/640
    • YUN00001_3600 @ 32/640
    • vis_png @ 32/640
      • 20160122_YUN00002_700_2500 @ vis_png/32/640
      • 20160122_YUN00020_2000_3800 @ vis_png/32/640
      • YUN00001_3600 @ vis_png/32/640
• 640_selective
  • 4__non_aug @ 640_selective
    • 2 @ 4__non_aug/640_selective
    • 2__0_14 @ 4__non_aug/640_selective
    • 10 @ 4__non_aug/640_selective
    • 10__rt3 @ 4__non_aug/640_selective
    • 10__0_14 @ 4__non_aug/640_selective
    • 100 @ 4__non_aug/640_selective
    • 100__0_14 @ 4__non_aug/640_selective
    • 1000 @ 4__non_aug/640_selective
    • 1000__0_14 @ 4__non_aug/640_selective
  • 4 @ 640_selective
    • 2 @ 4/640_selective
    • 5 @ 4/640_selective
    • 10 @ 4/640_selective
      • rt @ 10/4/640_selective
    • 100 @ 4/640_selective
    • vis @ 4/640_selective
    • no_aug @ 4/640_selective
      • stitched @ no_aug/4/640_selective
    • 1K @ 4/640_selective
    • 5K @ 4/640_selective
  • 16 @ 640_selective
    • 100 @ 16/640_selective
  • 32 @ 640_selective
    • 100 @ 32/640_selective

python visualizeDataset.py
--images="/home/abhineet/N/Datasets/617/Training_256_25_100/images/"
--annotations="/home/abhineet/N/Datasets/617/Training_256_25_100/labels/"
--n_classes=3

640x320 / misc

KERAS_BACKEND=theano THEANO_FLAGS=device=gpu,floatX=float32 python img_keras_train.py --save_weights_path=weights/ex1 --train_images="/data/617/images/training_320_640_25_100/images/" --train_annotations="/data/617/images/training_320_640_25_100/labels/" --val_images="/data/617/images/training_320_640_100_200/images/" --val_annotations="/data/617/images/training_320_640_100_200/labels/" --n_classes=3 --input_height=320 --input_width=640 --model_name="vgg_unet2"

--model_name="fcn32" --model_name="vgg_segnet"
--model_name="vgg_unet2"

KERAS_BACKEND=theano THEANO_FLAGS=device=gpu,floatX=float32 python img_keras_train.py --save_weights_path=weights/ex1 --train_images="/data/617/images/dataset1/images_prepped_train/" --train_annotations="/data/617/images/dataset1/annotations_prepped_train/" --val_images="/data/617/images/dataset1/images_prepped_test/" --val_annotations="/data/617/images/dataset1/annotations_prepped_test/" --n_classes=10 --input_height=320 --input_width=640 --model_name="vgg_segnet"

change tf.concat(0, [[self._batch_size], [num_dim, -1], [input_shape[2]]]) to tf.concat([[self._batch_size], [num_dim, -1], [input_shape[2]]], 0) and change another tf.concat similarly (also within this file). You should be fine.

256

KERAS_BACKEND=theano THEANO_FLAGS

="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_train.py --save_weights_path=weights/vgg_unet2_256_25_100_rot_15_90_flip --train_images="/data/617/images/training_256_256_25_100_rot_15_90_flip/images/" --train_annotations="/data/617/images/training_256_256_25_100_rot_15_90_flip/labels/" --val_images="/data/617/images/training_256_256_100_200_rot_90_180_flip/images/" --val_annotations="/data/617/images/training_256_256_100_200_rot_90_180_flip/labels/" --n_classes=3 --input_height=256 --input_width=256 --model_name="vgg_unet2" --epochs=1000

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/cuda/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_train.py --save_weights_path=weights/fcn32_256_25_100_rot_15_90_flip --train_images="/data/617/images/training_256_256_25_100_rot_15_90_flip/images/" --train_annotations="/data/617/images/training_256_256_25_100_rot_15_90_flip/labels/" --val_images="/data/617/images/training_256_256_100_200_rot_90_180_flip/images/" --val_annotations="/data/617/images/training_256_256_100_200_rot_90_180_flip/labels/" --n_classes=3 --input_height=256 --input_width=256 --model_name="fcn32" --epochs=1000

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_train.py --save_weights_path=weights/vgg_segnet_256_25_100_rot_15_90_flip --train_images="/data/617/images/training_256_256_25_100_rot_15_90_flip/images/" --train_annotations="/data/617/images/training_256_256_25_100_rot_15_90_flip/labels/" --val_images="/data/617/images/training_256_256_100_200_flip/images/" --val_annotations="/data/617/images/training_256_256_100_200_flip/labels/" --n_classes=3 --input_height=256 --input_width=256 --model_name="vgg_segnet" --epochs=1000

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_train.py --save_weights_path=weights/fcn8_256_25_100_rot_15_90_flip --train_images="/data/617/images/training_256_256_25_100_rot_15_90_flip/images/" --train_annotations="/data/617/images/training_256_256_25_100_rot_15_90_flip/labels/" --val_images="/data/617/images/training_256_256_100_200_flip/images/" --val_annotations="/data/617/images/training_256_256_100_200_flip/labels/" --n_classes=3 --input_height=256 --input_width=256 --model_name="fcn8" --epochs=1000

evaluation @ 256

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/cuda/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_predict.py --save_weights_path=weights/vgg_unet2_32_18_256_256_25_100_rot_15_125_235_345_flip/weights_490.h5 --test_images="/data/617/images/training_256_256_100_200_rot_90_180_flip/images/" --output_path="/data/617/images/vgg_unet2_max_val_acc_training_256_256_100_200_rot_90_180_flip/predictions/" --n_classes=3 --input_height=256 --input_width=256 --model_name="vgg_unet2"

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_predict.py --save_weights_path=weights/vgg_segnet_32_18_256_256_25_100_rot_15_125_235_345_flip/weights_266.h5 --test_images="/data/617/images/training_32_49_256_256_25_100_rot_15_125_235_345_flip/images/" --output_path="/data/617/images/vgg_segnet_max_mean_acc_training_32_49_256_256_25_100_rot_15_125_235_345_flip/predictions/" --n_classes=3 --input_height=256 --input_width=256 --model_name="vgg_segnet"

vis @ evaluation/256

python3 ../visDataset.py --images_path=/data/617/images/training_32_49_256_256_25_100_rot_15_125_235_345_flip/images --labels_path=/data/617/images/training_32_49_256_256_25_100_rot_15_125_235_345_flip/labels --seg_path=/data/617/images/vgg_segnet_max_mean_acc_training_32_49_256_256_25_100_rot_15_125_235_345_flip/predictions/raw --save_path=/data/617/images/vgg_segnet_max_mean_acc_training_32_49_256_256_25_100_rot_15_125_235_345_flip/vis --n_classes=3 --start_id=0 --end_id=-1 --save_stitched=0 --normalize_labels=1

hml @ evaluation/256

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/targets/x86_64-linux/include/,dnn.library_path=/usr/local/cuda-9.0/targets/x86_64-linux/lib/" python2 img_keras_predict.py --save_weights_path=weights/vgg_unet2_32_18_256_256_25_100_rot_15_125_235_345_flip/weights_490.h5 --test_images="/data/617/images/training_256_256_100_200_rot_90_180_flip/images/" --output_path="/data/617/images/vgg_unet2_max_val_acc_training_256_256_100_200_rot_90_180_flip/predictions/" --n_classes=3 --input_height=256 --input_width=256 --model_name="vgg_unet2"

zr vgg_unet2_32_18_256_256_25_100_rot_15_125_235_345_flip_weights_490 weights/vgg_unet2_32_18_256_256_25_100_rot_15_125_235_345_flip/weights_490.h5 zr img_seg_keras_data data

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/targets/x86_64-linux/include/,dnn.library_path=/usr/local/cuda-9.0/targets/x86_64-linux/lib/" python2 img_keras_predict.py --save_weights_path=weights/vgg_unet2_32_18_256_256_25_100_rot_15_125_235_345_flip/weights_490.h5 --test_images="/data/617/images/training_32_49_256_256_25_100_rot_15_125_235_345_flip/images/" --output_path="/data/617/images/vgg_unet2_max_val_acc_training_32_49_256_256_25_100_rot_15_125_235_345_flip/predictions/" --n_classes=3 --input_height=256 --input_width=256 --model_name="vgg_unet2"

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/targets/x86_64-linux/include/,dnn.library_path=/usr/local/cuda-9.0/targets/x86_64-linux/lib/" python2 img_keras_predict.py --save_weights_path=weights/vgg_unet2_32_18_256_256_25_100_rot_15_125_235_345_flip/weights_490.h5 --test_images="/data/617/images/training_32_49_256_256_25_100_rot_15_125_235_345_flip/images/" --output_path="/data/617/images/vgg_segnet_max_val_acc_training_32_49_256_256_25_100_rot_15_125_235_345_flip/predictions/" --n_classes=3 --input_height=256 --input_width=256 --model_name="vgg_segnet"

vis @ 256

hml @ vis/256

python3 visDataset.py --images_path=/data/617/images/training_256_256_100_200_rot_90_180_flip/images --labels_path=/data/617/images/training_256_256_100_200_rot_90_180_flip/labels --seg_path=/data/617/images/vgg_unet2_max_val_acc_training_256_256_100_200_rot_90_180_flip/predictions/raw --save_path=/data/617/images/vgg_unet2_max_val_acc_training_256_256_100_200_rot_90_180_flip/vis --n_classes=3 --start_id=0 --end_id=-1

python3 visDataset.py --images_path=/data/617/images/training_32_49_256_256_25_100_rot_15_125_235_345_flip/images --labels_path=/data/617/images/training_32_49_256_256_25_100_rot_15_125_235_345_flip/labels --seg_path=/data/617/images/vgg_unet2_max_val_acc_training_32_49_256_256_25_100_rot_15_125_235_345_flip/predictions/raw --save_path=/data/617/images/vgg_unet2_max_val_acc_training_32_49_256_256_25_100_rot_15_125_235_345_flip/vis --n_classes=3 --start_id=0 --end_id=-1

validation @ 256

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/cuda/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_unet2_32_18_256_256_25_100_rot_15_125_235_345_flip/weights_490.h5 --test_images="/data/617/images/validation_0_20_256_256_256_256/images/" --output_path="/data/617/images/vgg_unet2_max_val_acc_validation_0_20_256_256_256_256/predictions/" --n_classes=3 --input_height=256 --input_width=256 --model_name="vgg_unet2"

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/cuda/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/fcn32_32_18_256_256_25_100_rot_15_125_235_345_flip/weights_174.h5 --test_images="/data/617/images/validation_0_20_256_256_256_256/images/" --output_path="/data/617/images/fcn32_max_mean_acc_validation_0_20_256_256_256_256/predictions/" --n_classes=3 --input_height=256 --input_width=256 --model_name="fcn32"

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_segnet_32_18_256_256_25_100_rot_15_125_235_345_flip/weights_685.h5 --test_images="/data/617/images/validation_0_20_256_256_256_256/images/" --output_path="/data/617/images/validation_0_20_256_256_256_256/predictions/" --n_classes=3 --input_height=256 --input_width=256 --model_name="vgg_segnet"

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/fcn8_256_25_100_rot_15_90_flip/weights_685.h5 --test_images="/data/617/images/validation_0_20_256_256_256_256/images/" --output_path="/data/617/images/validation_0_20_256_256_256_256/predictions/" --n_classes=3 --input_height=256 --input_width=256 --model_name="fcn8"

videos @ 256

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/cuda/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_unet2_32_18_256_256_25_100_rot_15_125_235_345_flip/weights_490.h5 --test_images="/data/617/images/YUN00001_0_239_256_256_256_256/images/" --output_path="/data/617/images/YUN00001_0_239_256_256_256_256/vgg_unet2_32_18_256_256_25_100_rot_15_125_235_345_flip/" --n_classes=3 --input_height=256 --input_width=256 --model_name="vgg_unet2"

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_segnet_32_18_256_256_25_100_rot_15_125_235_345_flip/weights_685.h5 --test_images="/data/617/images/YUN00001_0_239_256_256_256_256/images/" --output_path="/data/617/images/YUN00001_0_239_256_256_256_256/vgg_segnet_32_18_256_256_25_100_rot_15_125_235_345_flip/" --n_classes=3 --input_height=256 --input_width=256 --model_name="vgg_segnet"

new @ 256

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/cuda/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_train.py --save_weights_path=weights/vgg_unet2_32_18_256_256_25_100_rot_15_125_235_345_flip --train_images="/data/617/images/training_0_31_256_256_25_100_rot_15_125_235_345_flip/images/" --train_annotations="/data/617/images/training_0_31_256_256_25_100_rot_15_125_235_345_flip/labels/" --val_images="/data/617/images/training_32_49_256_256_25_100_rot_15_125_235_345_flip/images/" --val_annotations="/data/617/images/training_32_49_256_256_25_100_rot_15_125_235_345_flip/labels/" --n_classes=3 --input_height=256 --input_width=256 --model_name="vgg_unet2" --epochs=1000

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/cuda/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_train.py --save_weights_path=weights/fcn32_32_18_256_256_25_100_rot_15_125_235_345_flip --train_images="/data/617/images/training_0_31_256_256_25_100_rot_15_125_235_345_flip/images/" --train_annotations="/data/617/images/training_0_31_256_256_25_100_rot_15_125_235_345_flip/labels/" --val_images="/data/617/images/training_32_49_256_256_25_100_rot_15_125_235_345_flip/images/" --val_annotations="/data/617/images/training_32_49_256_256_25_100_rot_15_125_235_345_flip/labels/" --n_classes=3 --input_height=256 --input_width=256 --model_name="fcn32" --epochs=1000

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_train.py --save_weights_path=weights/vgg_segnet_32_18_256_256_25_100_rot_15_125_235_345_flip --train_images="/data/617/images/training_0_31_256_256_25_100_rot_15_125_235_345_flip/images/" --train_annotations="/data/617/images/training_0_31_256_256_25_100_rot_15_125_235_345_flip/labels/" --val_images="/data/617/images/training_32_49_256_256_25_100_rot_15_125_235_345_flip/images/" --val_annotations="/data/617/images/training_32_49_256_256_25_100_rot_15_125_235_345_flip/labels/" --n_classes=3 --input_height=256 --input_width=256 --model_name="vgg_segnet" --epochs=1000

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_train.py --save_weights_path=weights/fcn8_32_18_256_256_25_100_rot_15_125_235_345_flip --train_images="/data/617/images/training_0_31_256_256_25_100_rot_15_125_235_345_flip/images/" --train_annotations="/data/617/images/training_0_31_256_256_25_100_rot_15_125_235_345_flip/labels/" --val_images="/data/617/images/training_32_49_256_256_25_100_rot_15_125_235_345_flip/images/" --val_annotations="/data/617/images/training_32_49_256_256_25_100_rot_15_125_235_345_flip/labels/" --n_classes=3 --input_height=256 --input_width=256 --model_name="fcn8" --epochs=1000

384

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/cuda/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_train.py --save_weights_path=weights/vgg_unet2_0_31_384_384_25_100_rot_15_345_4_flip --train_images="/data/617/images/training_0_31_384_384_25_100_rot_15_345_4_flip/images/" --train_annotations="/data/617/images/training_0_31_384_384_25_100_rot_15_345_4_flip/labels/" --val_images="/data/617/images/training_32_49_384_384_25_100_rot_15_345_4_flip/images/" --val_annotations="/data/617/images/training_32_49_384_384_25_100_rot_15_345_4_flip/labels/" --n_classes=3 --input_height=384 --input_width=384 --model_name="vgg_unet2" --epochs=1000

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/cuda/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_train.py --save_weights_path=weights/fcn32_0_31_384_384_25_100_rot_15_345_4_flip --train_images="/data/617/images/training_0_31_384_384_25_100_rot_15_345_4_flip/images/" --train_annotations="/data/617/images/training_0_31_384_384_25_100_rot_15_345_4_flip/labels/" --val_images="/data/617/images/training_32_49_384_384_25_100_rot_15_345_4_flip/images/" --val_annotations="/data/617/images/training_32_49_384_384_25_100_rot_15_345_4_flip/labels/" --n_classes=3 --input_height=384 --input_width=384 --model_name="fcn32" --epochs=1000

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_train.py --save_weights_path=weights/vgg_segnet_0_31_384_384_25_100_rot_15_345_4_flip --train_images="/data/617/images/training_0_31_384_384_25_100_rot_15_345_4_flip/images/" --train_annotations="/data/617/images/training_0_31_384_384_25_100_rot_15_345_4_flip/labels/" --val_images="/data/617/images/training_32_49_384_384_25_100_rot_15_345_4_flip/images/" --val_annotations="/data/617/images/training_32_49_384_384_25_100_rot_15_345_4_flip/labels/" --n_classes=3 --input_height=384 --input_width=384 --model_name="vgg_segnet" --epochs=1000

evaluation @ 384

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_predict.py --save_weights_path=weights/vgg_unet2_32_18_256_256_25_100_rot_15_125_235_345_flip/weights_490.h5 --test_images="/data/617/images/training_32_49_384_384_25_100_rot_15_345_4_flip/images/" --output_path="/data/617/images/vgg_unet2_max_val_acc_training_32_49_384_384_25_100_rot_15_345_4_flip/predictions/" --n_classes=3 --input_height=256 --input_width=256 --model_name="vgg_unet2"

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_predict.py --save_weights_path=weights/vgg_segnet_0_31_384_384_25_100_rot_15_345_4_flip/weights_583.h5 --test_images="/data/617/images/training_32_49_384_384_25_100_rot_15_345_4_flip/images/" --output_path="/data/617/images/vgg_segnet_max_val_acc_training_32_49_384_384_25_100_rot_15_345_4_flip/predictions/" --n_classes=3 --input_height=384 --input_width=384 --model_name="vgg_segnet"

vis @ evaluation/384

python3 ../visDataset.py --images_path=/data/617/images/training_32_49_384_384_25_100_rot_15_345_4_flip/images --labels_path=/data/617/images/training_32_49_384_384_25_100_rot_15_345_4_flip/labels --seg_path=/data/617/images/vgg_segnet_max_val_acc_training_32_49_384_384_25_100_rot_15_345_4_flip/predictions/raw --save_path=/data/617/images/vgg_segnet_max_val_acc_training_32_49_384_384_25_100_rot_15_345_4_flip/vis --n_classes=3 --start_id=0 --end_id=-1 --save_stitched=0 --normalize_labels=1

hml @ evaluation/384

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/targets/x86_64-linux/include/,dnn.library_path=/usr/local/cuda-9.0/targets/x86_64-linux/lib/" python2 img_keras_predict.py --save_weights_path=weights/vgg_unet2_0_31_384_384_25_100_rot_15_345_4_flip/weights_497.h5 --test_images="/data/617/images/training_32_49_384_384_25_100_rot_15_345_4_flip/images/" --output_path="/data/617/images/vgg_unet2_max_val_acc_training_32_49_384_384_25_100_rot_15_345_4_flip/predictions/" --n_classes=3 --input_height=384 --input_width=384 --model_name="vgg_unet2"

validation @ 384

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/cuda/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_unet2_32_18_384_384_25_100_rot_15_125_235_345_flip/weights_490.h5 --test_images="/data/617/images/validation_0_20_384_384_384_384/images/" --output_path="/data/617/images/vgg_unet2_max_val_acc_validation_0_20_384_384_384_384/predictions/" --n_classes=3 --input_height=384 --input_width=384 --model_name="vgg_unet2"

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/cuda/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/fcn32_32_18_384_384_25_100_rot_15_125_235_345_flip/weights_174.h5 --test_images="/data/617/images/validation_0_20_384_384_384_384/images/" --output_path="/data/617/images/fcn32_max_mean_acc_validation_0_20_384_384_384_384/predictions/" --n_classes=3 --input_height=384 --input_width=384 --model_name="fcn32"

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_segnet_32_18_384_384_25_100_rot_15_125_235_345_flip/weights_685.h5 --test_images="/data/617/images/validation_0_20_384_384_384_384/images/" --output_path="/data/617/images/validation_0_20_384_384_384_384/predictions/" --n_classes=3 --input_height=384 --input_width=384 --model_name="vgg_segnet"

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/fcn8_384_25_100_rot_15_90_flip/weights_685.h5 --test_images="/data/617/images/validation_0_20_384_384_384_384/images/" --output_path="/data/617/images/validation_0_20_384_384_384_384/predictions/" --n_classes=3 --input_height=384 --input_width=384 --model_name="fcn8"

videos @ 384

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/cuda/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_unet2_32_18_384_384_25_100_rot_15_125_235_345_flip/weights_490.h5 --test_images="/data/617/images/YUN00001_0_239_384_384_384_384/images/" --output_path="/data/617/images/YUN00001_0_239_384_384_384_384/vgg_unet2_32_18_384_384_25_100_rot_15_125_235_345_flip/" --n_classes=3 --input_height=384 --input_width=384 --model_name="vgg_unet2"

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_segnet_32_18_384_384_25_100_rot_15_125_235_345_flip/weights_685.h5 --test_images="/data/617/images/YUN00001_0_239_384_384_384_384/images/" --output_path="/data/617/images/YUN00001_0_239_384_384_384_384/vgg_segnet_32_18_384_384_25_100_rot_15_125_235_345_flip/" --n_classes=3 --input_height=384 --input_width=384 --model_name="vgg_segnet"

512

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_train.py --save_weights_path=weights/vgg_unet2_0_31_512_512_25_100_rot_15_345_4_flip --train_images="/data/617/images/training_0_31_512_512_25_100_rot_15_345_4_flip/images/" --train_annotations="/data/617/images/training_0_31_512_512_25_100_rot_15_345_4_flip/labels/" --val_images="/data/617/images/training_32_49_512_512_25_100_rot_15_345_4_flip/images/" --val_annotations="/data/617/images/training_32_49_512_512_25_100_rot_15_345_4_flip/labels/" --n_classes=3 --input_height=512 --input_width=512 --model_name="vgg_unet2" --epochs=1000

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/cuda/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_train.py --save_weights_path=weights/fcn32_0_31_512_512_25_100_rot_15_345_4_flip --train_images="/data/617/images/training_0_31_512_512_25_100_rot_15_345_4_flip/images/" --train_annotations="/data/617/images/training_0_31_512_512_25_100_rot_15_345_4_flip/labels/" --val_images="/data/617/images/training_32_49_512_512_25_100_rot_15_345_4_flip/images/" --val_annotations="/data/617/images/training_32_49_512_512_25_100_rot_15_345_4_flip/labels/" --n_classes=3 --input_height=512 --input_width=512 --model_name="fcn32" --epochs=1000

evaluation @ 512

hml @ evaluation/512

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/targets/x86_64-linux/include/,dnn.library_path=/usr/local/cuda-9.0/targets/x86_64-linux/lib/" python2 img_keras_predict.py --save_weights_path=weights/vgg_unet2_0_31_512_512_25_100_rot_15_345_4_flip/weights_max_val_acc_140.h5 --test_images="/data/617/images/training_32_49_512_512_25_100_rot_15_345_4_flip/images/" --output_path="/data/617/images/vgg_unet2_max_val_acc_training_32_49_512_512_25_100_rot_15_345_4_flip/predictions/" --n_classes=3 --input_height=512 --input_width=512 --model_name="vgg_unet2"

vis @ 512

hml @ vis/512

python3 ../visDataset.py --images_path=/data/617/images/training_32_49_512_512_25_100_rot_15_345_4_flip/images --labels_path=/data/617/images/training_32_49_512_512_25_100_rot_15_345_4_flip/labels --seg_path=/data/617/images/vgg_unet2_max_val_acc_training_32_49_512_512_25_100_rot_15_345_4_flip/predictions/raw --save_path=/data/617/images/vgg_unet2_max_val_acc_training_32_49_512_512_25_100_rot_15_345_4_flip/vis --n_classes=3 --start_id=0 --end_id=-1

validation @ 512

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/targets/x86_64-linux/include/,dnn.library_path=/usr/local/cuda-9.0/targets/x86_64-linux/lib/" python2 img_keras_predict.py --save_weights_path=weights/vgg_unet2_0_31_512_512_25_100_rot_15_345_4_flip/weights_max_val_acc_140.h5 --test_images="/data/617/images/validation_0_20_512_512_512_512/images/" --output_path="/data/617/images/vgg_unet2_max_val_acc_validation_0_20_512_512_512_512/predictions/" --n_classes=3 --input_height=512 --input_width=512 --model_name="vgg_unet2"

stitching @ validation/512

python3 ../stitchSubPatchDataset.py src_path=/data/617/images/validation/images patch_seq_path=/data/617/images/vgg_unet2_max_val_acc_validation_0_20_512_512_512_512/predictions/raw stitched_seq_path=/data/617/images/vgg_unet2_max_val_acc_validation_0_20_512_512_512_512/predictions/stitched patch_height=512 patch_width=512 start_id=0 end_id=20 show_img=0 stacked=1 method=1 normalize_patches=1

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/cuda/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_unet2_32_18_512_512_25_100_rot_15_125_235_345_flip/weights_490.h5 --test_images="/data/617/images/validation_0_20_512_512_512_512/images/" --output_path="/data/617/images/vgg_unet2_max_val_acc_validation_0_20_512_512_512_512/predictions/" --n_classes=3 --input_height=512 --input_width=512 --model_name="vgg_unet2"

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/cuda/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/fcn32_32_18_512_512_25_100_rot_15_125_235_345_flip/weights_174.h5 --test_images="/data/617/images/validation_0_20_512_512_512_512/images/" --output_path="/data/617/images/fcn32_max_mean_acc_validation_0_20_512_512_512_512/predictions/" --n_classes=3 --input_height=512 --input_width=512 --model_name="fcn32"

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/fcn8_512_25_100_rot_15_90_flip/weights_685.h5 --test_images="/data/617/images/validation_0_20_512_512_512_512/images/" --output_path="/data/617/images/validation_0_20_512_512_512_512/predictions/" --n_classes=3 --input_height=512 --input_width=512 --model_name="fcn8"

videos @ 512

KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/cuda/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_unet2_32_18_512_512_25_100_rot_15_125_235_345_flip/weights_490.h5 --test_images="/data/617/images/YUN00001_0_239_512_512_512_512/images/" --output_path="/data/617/images/YUN00001_0_239_512_512_512_512/vgg_unet2_32_18_512_512_25_100_rot_15_125_235_345_flip/" --n_classes=3 --input_height=512 --input_width=512 --model_name="vgg_unet2"

640

4__non_aug @ 640

CUDA_VISIBLE_DEVICES=0 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_train.py --save_weights_path=weights/vgg_unet2_0_3_640_640_640_640 --train_images="/data/617/images/training_0_3_640_640_640_640/images/" --train_annotations="/data/617/images/training_0_3_640_640_640_640/labels/" --val_images="/data/617/images/training_32_49_640_640_640_640/images/" --val_annotations="/data/617/images/training_32_49_640_640_640_640/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000

4 @ 640

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_train.py --save_weights_path=weights/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip --train_images="/data/617/images/training_0_3_640_640_64_256_rot_15_345_4_flip/images/" --train_annotations="/data/617/images/training_0_3_640_640_64_256_rot_15_345_4_flip/labels/" --val_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --val_annotations="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000

evaluation @ 4/640

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip/weights_max_val_acc --test_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --output_path="log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_64_256_rot_15_345_4_flip_max_val_acc/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2"

python3 ../visDataset.py --images_path=/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images --labels_path=/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels --seg_path=log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_64_256_rot_15_345_4_flip_max_val_acc/raw --save_path=log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_64_256_rot_15_345_4_flip_max_val_acc/vis --n_classes=3 --start_id=0 --end_id=-1

4_49 @ 4/640

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip/weights_max_val_acc --test_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --output_path="log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_64_256_rot_15_345_4_flip_max_val_acc/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2"

python3 ../visDataset.py --images_path=/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images --labels_path=/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels --seg_path=log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_64_256_rot_15_345_4_flip_max_val_acc/raw --save_path=log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_64_256_rot_15_345_4_flip_max_val_acc/vis --n_classes=3 --start_id=0 --end_id=-1

no_aug @ 4/640

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip/weights_max_val_acc --test_images="/data/617/images/training_4_49_640_640_640_640/images/" --output_path="log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip/training_4_49_640_640_640_640_max_val_acc/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2"

python3 ../visDataset.py --images_path=/data/617/images/training_32_49_640_640_640_640/images --labels_path=/data/617/images/training_4_49_640_640_640_640/labels --seg_path=log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_640_640_max_val_acc/raw --save_path=log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip/training_4_49_640_640_640_640_max_val_acc/vis --n_classes=3 --start_id=0 --end_id=-1

stitched @ no_aug/4/640

python3 ../stitchSubPatchDataset.py src_path=/data/617/images/training_32_49/images img_ext=jpg patch_seq_path=log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_640_640_max_val_acc/raw stitched_seq_path=log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip/training_32_49_max_val_acc/raw patch_height=640 patch_width=640 start_id=0 end_id=-1 show_img=0 stacked=0 method=1 normalize_patches=0 img_ext=png

python3 ../visDataset.py --images_path=/data/617/images/training_32_49/images --labels_path=/data/617/images/training_32_49/labels --seg_path=log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip/training_32_49_max_val_acc/raw --save_path=log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip/training_32_49_max_val_acc/vis --n_classes=3 --start_id=0 --end_id=-1

no_aug__4_49 @ 4/640

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip/weights_max_val_acc --test_images="/data/617/images/training_4_49_640_640_640_640/images/" --output_path="log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_640_640_max_val_acc/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2"

python3 ../stitchSubPatchDataset.py src_path=/data/617/images/training_4_49/images labels_path=/data/617/images/training_4_49/labels img_ext=jpg patch_seq_path=log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip/training_4_49_640_640_640_640_max_val_acc/raw stitched_seq_path=log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip/training_4_49_max_val_acc/raw patch_height=640 patch_width=640 start_id=0 end_id=-1 show_img=0 stacked=0 method=1 normalize_patches=0 img_ext=png

8 @ 640

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_train.py --save_weights_path=weights/vgg_unet2_0_7_640_640_64_256_rot_15_345_4_flip --train_images="/data/617/images/training_0_7_640_640_64_256_rot_15_345_4_flip/images/" --train_annotations="/data/617/images/training_0_7_640_640_64_256_rot_15_345_4_flip/labels/" --val_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --val_annotations="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000 --load_weights=1

evaluation @ 8/640

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_unet2_0_7_640_640_64_256_rot_15_345_4_flip/weights_max_val_acc --test_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --output_path="log/vgg_unet2_0_7_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_64_256_rot_15_345_4_flip_max_val_acc/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2"

python3 ../visDataset.py --images_path=/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images --labels_path=/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels --seg_path=log/vgg_unet2_0_7_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_64_256_rot_15_345_4_flip_max_val_acc/raw --save_path=log/vgg_unet2_0_7_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_64_256_rot_15_345_4_flip_max_val_acc/vis --n_classes=3 --start_id=0 --end_id=-1

no_aug @ 8/640

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_unet2_0_7_640_640_64_256_rot_15_345_4_flip/weights_max_val_acc --test_images="/data/617/images/training_32_49_640_640_640_640/images/" --output_path="log/vgg_unet2_0_7_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_640_640_max_val_acc/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2"

python3 ../visDataset.py --images_path=/data/617/images/training_32_49_640_640_640_640/images --labels_path=/data/617/images/training_32_49_640_640_640_640/labels --seg_path=log/vgg_unet2_0_7_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_640_640_max_val_acc/raw --save_path=log/vgg_unet2_0_7_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_640_640_max_val_acc/vis --n_classes=3 --start_id=0 --end_id=-1

stitched @ no_aug/8/640

python3 ../stitchSubPatchDataset.py src_path=/data/617/images/training_32_49/images img_ext=jpg patch_seq_path=log/vgg_unet2_0_7_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_640_640_max_val_acc/raw stitched_seq_path=log/vgg_unet2_0_7_640_640_64_256_rot_15_345_4_flip/training_32_49_max_val_acc/raw patch_height=640 patch_width=640 start_id=0 end_id=-1 show_img=0 stacked=0 method=1 normalize_patches=0 img_ext=png

python3 ../visDataset.py --images_path=/data/617/images/training_32_49/images --labels_path=/data/617/images/training_32_49/labels --seg_path=log/vgg_unet2_0_7_640_640_64_256_rot_15_345_4_flip/training_32_49_max_val_acc/raw --save_path=log/vgg_unet2_0_7_640_640_64_256_rot_15_345_4_flip/training_32_49_max_val_acc/vis --n_classes=3 --start_id=0 --end_id=-1

16 @ 640

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_train.py --save_weights_path=weights/vgg_unet2_0_15_640_640_64_256_rot_15_345_4_flip --train_images="/data/617/images/training_0_15_640_640_64_256_rot_15_345_4_flip/images/" --train_annotations="/data/617/images/training_0_15_640_640_64_256_rot_15_345_4_flip/labels/" --val_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --val_annotations="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000

continue_133 @ 16/640

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_train.py --save_weights_path=weights/vgg_unet2_0_15_640_640_64_256_rot_15_345_4_flip --train_images="/data/617/images/training_0_15_640_640_64_256_rot_15_345_4_flip/images/" --train_annotations="/data/617/images/training_0_15_640_640_64_256_rot_15_345_4_flip/labels/" --val_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --val_annotations="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000 --load_weights=weights/vgg_unet2_0_15_640_640_64_256_rot_15_345_4_flip/weights_max_acc_133.h5

continue_latest @ 16/640

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_train.py --save_weights_path=weights/vgg_unet2_0_15_640_640_64_256_rot_15_345_4_flip --train_images="/data/617/images/training_0_15_640_640_64_256_rot_15_345_4_flip/images/" --train_annotations="/data/617/images/training_0_15_640_640_64_256_rot_15_345_4_flip/labels/" --val_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --val_annotations="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000 --load_weights=1

evaluation @ 16/640

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_unet2_0_15_640_640_64_256_rot_15_345_4_flip/weights_max_val_acc --test_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --output_path="log/vgg_unet2_0_15_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_64_256_rot_15_345_4_flip_max_val_acc/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2"

python3 ../visDataset.py --images_path=/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images --labels_path=/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels --seg_path=log/vgg_unet2_0_15_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_64_256_rot_15_345_4_flip_max_val_acc/raw --save_path=log/vgg_unet2_0_15_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_64_256_rot_15_345_4_flip_max_val_acc/vis --n_classes=3 --start_id=0 --end_id=-1

no_aug @ 16/640

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_unet2_0_15_640_640_64_256_rot_15_345_4_flip/weights_max_val_acc --test_images="/data/617/images/training_32_49_640_640_640_640/images/" --output_path="log/vgg_unet2_0_15_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_640_640_max_val_acc/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2"

python3 ../visDataset.py --images_path=/data/617/images/training_32_49_640_640_640_640/images --labels_path=/data/617/images/training_32_49_640_640_640_640/labels --seg_path=log/vgg_unet2_0_15_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_640_640_max_val_acc/raw --save_path=log/vgg_unet2_0_15_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_640_640_max_val_acc/vis --n_classes=3 --start_id=0 --end_id=-1

stitched @ no_aug/16/640

python3 ../stitchSubPatchDataset.py src_path=/data/617/images/training_32_49/images img_ext=jpg patch_seq_path=log/vgg_unet2_0_15_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_640_640_max_val_acc/raw stitched_seq_path=log/vgg_unet2_0_15_640_640_64_256_rot_15_345_4_flip/training_32_49_max_val_acc/raw patch_height=640 patch_width=640 start_id=0 end_id=-1 show_img=0 stacked=0 method=1 normalize_patches=0 img_ext=png

python3 ../visDataset.py --images_path=/data/617/images/training_32_49/images --labels_path=/data/617/images/training_32_49/labels --seg_path=log/vgg_unet2_0_15_640_640_64_256_rot_15_345_4_flip/training_32_49_max_val_acc/raw --save_path=log/vgg_unet2_0_15_640_640_64_256_rot_15_345_4_flip/training_32_49_max_val_acc/vis --n_classes=3 --start_id=0 --end_id=-1

24 @ 640

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_train.py --save_weights_path=weights/vgg_unet2_0_23_640_640_64_256_rot_15_345_4_flip --train_images="/data/617/images/training_0_23_640_640_64_256_rot_15_345_4_flip/images/" --train_annotations="/data/617/images/training_0_23_640_640_64_256_rot_15_345_4_flip/labels/" --val_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --val_annotations="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000

continue_171 @ 24/640

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_train.py --save_weights_path=weights/vgg_unet2_0_23_640_640_64_256_rot_15_345_4_flip --train_images="/data/617/images/training_0_23_640_640_64_256_rot_15_345_4_flip/images/" --train_annotations="/data/617/images/training_0_23_640_640_64_256_rot_15_345_4_flip/labels/" --val_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --val_annotations="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000 --load_weights=weights/vgg_unet2_0_23_640_640_64_256_rot_15_345_4_flip/weights_max_acc_171.h5

continue_latest @ 24/640

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_train.py --save_weights_path=weights/vgg_unet2_0_23_640_640_64_256_rot_15_345_4_flip --train_images="/data/617/images/training_0_23_640_640_64_256_rot_15_345_4_flip/images/" --train_annotations="/data/617/images/training_0_23_640_640_64_256_rot_15_345_4_flip/labels/" --val_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --val_annotations="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000 --load_weights=1

evaluation @ 24/640

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_unet2_0_23_640_640_64_256_rot_15_345_4_flip/weights_max_val_acc --test_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --output_path="log/vgg_unet2_0_23_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_64_256_rot_15_345_4_flip_max_val_acc/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2"

python3 ../visDataset.py --images_path=/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images --labels_path=/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels --seg_path=log/vgg_unet2_0_23_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_64_256_rot_15_345_4_flip_max_val_acc/raw --save_path=log/vgg_unet2_0_23_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_64_256_rot_15_345_4_flip_max_val_acc/vis --n_classes=3 --start_id=0 --end_id=-1

no_aug @ 24/640

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_unet2_0_23_640_640_64_256_rot_15_345_4_flip/weights_max_val_acc --test_images="/data/617/images/training_32_49_640_640_640_640/images/" --output_path="log/vgg_unet2_0_23_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_640_640_max_val_acc/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2"

python3 ../visDataset.py --images_path=/data/617/images/training_32_49_640_640_640_640/images --labels_path=/data/617/images/training_32_49_640_640_640_640/labels --seg_path=log/vgg_unet2_0_23_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_640_640_max_val_acc/raw --save_path=log/vgg_unet2_0_23_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_640_640_max_val_acc/vis --n_classes=3 --start_id=0 --end_id=-1

stitched @ no_aug/24/640

python3 ../stitchSubPatchDataset.py src_path=/data/617/images/training_32_49/images img_ext=jpg patch_seq_path=log/vgg_unet2_0_23_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_640_640_max_val_acc/raw stitched_seq_path=log/vgg_unet2_0_23_640_640_64_256_rot_15_345_4_flip/training_32_49_max_val_acc/raw patch_height=640 patch_width=640 start_id=0 end_id=-1 show_img=0 stacked=0 method=1 normalize_patches=0 img_ext=png

python3 ../visDataset.py --images_path=/data/617/images/training_32_49/images --labels_path=/data/617/images/training_32_49/labels --seg_path=log/vgg_unet2_0_23_640_640_64_256_rot_15_345_4_flip/training_32_49_max_val_acc/raw --save_path=log/vgg_unet2_0_23_640_640_64_256_rot_15_345_4_flip/training_32_49_max_val_acc/vis --n_classes=3 --start_id=0 --end_id=-1

32 @ 640

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_train.py --save_weights_path=weights/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip --train_images="/data/617/images/training_0_31_640_640_64_256_rot_15_345_4_flip/images/" --train_annotations="/data/617/images/training_0_31_640_640_64_256_rot_15_345_4_flip/labels/" --val_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --val_annotations="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000

evaluation @ 32/640

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/weights_max_val_acc_189.h5 --test_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --output_path="/data/617/images/vgg_unet2_max_val_acc_training_32_49_640_640_64_256_rot_15_345_4_flip/predictions/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2"

vis @ 32/640

python3 ../visDataset.py --images_path=/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images --labels_path=/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels --seg_path=/data/617/images/vgg_unet2_max_val_acc_training_32_49_640_640_64_256_rot_15_345_4_flip/predictions/raw --save_path=/data/617/images/vgg_unet2_max_val_acc_training_32_49_640_640_64_256_rot_15_345_4_flip/vis --n_classes=3 --start_id=0 --end_id=-1

no_aug @ 32/640

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/weights_max_val_acc --test_images="/data/617/images/training_32_49_640_640_640_640/images/" --output_path="log/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_640_640_max_val_acc/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2"

python3 ../visDataset.py --images_path=/data/617/images/training_32_49_640_640_640_640/images --labels_path=/data/617/images/training_32_49_640_640_640_640/labels --seg_path=log/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_640_640_max_val_acc/raw --save_path=log/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_640_640_max_val_acc/vis --n_classes=3 --start_id=0 --end_id=-1

stitched @ no_aug/32/640

python3 ../stitchSubPatchDataset.py src_path=/data/617/images/training_32_49/images img_ext=jpg patch_seq_path=log/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_640_640_max_val_acc/raw stitched_seq_path=log/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/training_32_49_max_val_acc/raw patch_height=640 patch_width=640 start_id=0 end_id=-1 show_img=0 stacked=0 method=1 normalize_patches=0 img_ext=png

python3 ../visDataset.py --images_path=/data/617/images/training_32_49/images --labels_path=/data/617/images/training_32_49/labels --seg_path=log/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/training_32_49_max_val_acc/raw --save_path=log/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/training_32_49_max_val_acc/vis --n_classes=3 --start_id=0 --end_id=-1

validation @ 32/640

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_predict.py --save_weights_path=weights/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/weights_max_val_acc_189.h5 --test_images="/data/617/images/validation_0_20_640_640_640_640/images/" --output_path="/data/617/images/vgg_unet2_max_val_acc_validation_0_20_640_640_640_640/predictions/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2"

stitching @ 32/640

python3 ../stitchSubPatchDataset.py src_path=/data/617/images/validation/images patch_seq_path=/data/617/images/vgg_unet2_max_val_acc_validation_0_20_640_640_640_640/predictions/raw stitched_seq_path=/data/617/images/vgg_unet2_max_val_acc_validation_0_20_640_640_640_640/predictions/stitched patch_height=640 patch_width=640 start_id=0 end_id=20 show_img=0 stacked=1 method=1 normalize_patches=1

zr vgg_unet2_max_val_acc_validation_0_20_640_640_640_640_stitched /data/617/images/vgg_unet2_max_val_acc_validation_0_20_640_640_640_640/predictions/stitched

YUN00001 @ 32/640

CUDA_VISIBLE_DEVICES=0 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/weights_max_val_acc --test_images="/data/617/images/YUN00001_0_8999_640_640_640_640/images/" --output_path="log/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/training_32_49_640_640_640_640_max_val_acc/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2"

python3 ../stitchSubPatchDataset.py src_path=/data/617/images/YUN00001/images patch_seq_path=log/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/YUN00001_0_8999_640_640_640_640_max_val_acc/raw stitched_seq_path=log/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/YUN00001_max_val_acc/raw patch_height=640 patch_width=640 start_id=0 end_id=-1 show_img=0 stacked=1 method=1 normalize_patches=0 img_ext=jpg out_ext=mkv width=1920 height=1080

YUN00001_3600 @ 32/640

CUDA_VISIBLE_DEVICES=0 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/weights_max_val_acc --test_images="/data/617/images/YUN00001_3600_0_3599_640_640_640_640/images/" --output_path="log/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/YUN00001_3600_0_3599_640_640_640_640_max_val_acc/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2"

python3 ../stitchSubPatchDataset.py src_path=/data/617/images/YUN00001_3600/images patch_seq_path=log/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/YUN00001_3600_0_3599_640_640_640_640_max_val_acc/raw stitched_seq_path=log/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/YUN00001_3600_max_val_acc/raw patch_height=640 patch_width=640 start_id=0 end_id=-1 show_img=0 stacked=1 method=1 normalize_patches=0 img_ext=jpg out_ext=mkv width=1920 height=1080

vis_png @ 32/640

20160122_YUN00002_700_2500 @ vis_png/32/640

python3 ../stitchSubPatchDataset.py src_path=/data/617/images/20160122_YUN00002_700_2500/images img_ext=jpg patch_seq_path=log/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/20160122_YUN00002_700_2500_0_1799_640_640_640_640_max_val_acc/raw stitched_seq_path=log/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/20160122_YUN00002_700_2500_max_val_acc patch_height=640 patch_width=640 start_id=0 end_id=-1 show_img=0 stacked=0 method=1 normalize_patches=0 img_ext=jpg out_ext=png

20160122_YUN00020_2000_3800 @ vis_png/32/640

python3 ../stitchSubPatchDataset.py src_path=/data/617/images/20160122_YUN00020_2000_3800/images img_ext=jpg patch_seq_path=log/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/20160122_YUN00020_2000_3800_0_1799_640_640_640_640_max_val_acc/raw stitched_seq_path=log/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/20160122_YUN00020_2000_3800_max_val_acc patch_height=640 patch_width=640 start_id=0 end_id=-1 show_img=0 stacked=0 method=1 normalize_patches=0 img_ext=jpg out_ext=png

YUN00001_3600 @ vis_png/32/640

python3 ../stitchSubPatchDataset.py src_path=/data/617/images/YUN00001_3600/images img_ext=jpg patch_seq_path=log/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/YUN00001_3600_0_3599_640_640_640_640_max_val_acc/raw stitched_seq_path=log/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip/YUN00001_3600_max_val_acc patch_height=640 patch_width=640 start_id=0 end_id=-1 show_img=0 stacked=0 method=1 normalize_patches=0 img_ext=jpg out_ext=png

640_selective

4__non_aug @ 640_selective

2 @ 4__non_aug/640_selective

CUDA_VISIBLE_DEVICES=0 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_train.py --save_weights_path=weights/vgg_unet2_0_3_640_640_640_640_2_rt2 --train_images="/data/617/images/training_0_3_640_640_640_640/images/" --train_annotations="/data/617/images/training_0_3_640_640_640_640/labels/" --val_images="/data/617/images/training_32_49_640_640_640_640/images/" --val_annotations="/data/617/images/training_32_49_640_640_640_640/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000 --selective_loss=2 --load_weights=1

2__0_14 @ 4__non_aug/640_selective

CUDA_VISIBLE_DEVICES=0 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_train.py --save_weights_path=weights/vgg_unet2_0_3_640_640_640_640_0_14_2 --train_images="/data/617/images/training_0_3_640_640_640_640/images/" --train_annotations="/data/617/images/training_0_3_640_640_640_640/labels/" --val_images="/data/617/images/training_32_49_640_640_640_640/images/" --val_annotations="/data/617/images/training_32_49_640_640_640_640/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000 --selective_loss=2 --load_weights=1 --start_id=0 --end_id=14

10 @ 4__non_aug/640_selective

CUDA_VISIBLE_DEVICES=0 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_train.py --save_weights_path=weights/vgg_unet2_0_3_640_640_640_640_10_rt2 --train_images="/data/617/images/training_0_3_640_640_640_640/images/" --train_annotations="/data/617/images/training_0_3_640_640_640_640/labels/" --val_images="/data/617/images/training_32_49_640_640_640_640/images/" --val_annotations="/data/617/images/training_32_49_640_640_640_640/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000 --selective_loss=10 --load_weights=0

10__rt3 @ 4__non_aug/640_selective

CUDA_VISIBLE_DEVICES=0 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_train.py --save_weights_path=weights/vgg_unet2_0_3_640_640_640_640_10_rt3 --train_images="/data/617/images/training_0_3_640_640_640_640/images/" --train_annotations="/data/617/images/training_0_3_640_640_640_640/labels/" --val_images="/data/617/images/training_32_49_640_640_640_640/images/" --val_annotations="/data/617/images/training_32_49_640_640_640_640/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000 --selective_loss=10 --load_weights=0

10__0_14 @ 4__non_aug/640_selective

CUDA_VISIBLE_DEVICES=0 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_train.py --save_weights_path=weights/vgg_unet2_0_3_640_640_640_640_0_14_10 --train_images="/data/617/images/training_0_3_640_640_640_640/images/" --train_annotations="/data/617/images/training_0_3_640_640_640_640/labels/" --val_images="/data/617/images/training_32_49_640_640_640_640/images/" --val_annotations="/data/617/images/training_32_49_640_640_640_640/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000 --selective_loss=10 --load_weights=1 --start_id=0 --end_id=14

100 @ 4__non_aug/640_selective

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_train.py --save_weights_path=weights/vgg_unet2_0_3_640_640_640_640_100_rt --train_images="/data/617/images/training_0_3_640_640_640_640/images/" --train_annotations="/data/617/images/training_0_3_640_640_640_640/labels/" --val_images="/data/617/images/training_32_49_640_640_640_640/images/" --val_annotations="/data/617/images/training_32_49_640_640_640_640/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000 --selective_loss=100 --load_weights=1

100__0_14 @ 4__non_aug/640_selective

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_train.py --save_weights_path=weights/vgg_unet2_0_3_640_640_640_640_0_14_100 --train_images="/data/617/images/training_0_3_640_640_640_640/images/" --train_annotations="/data/617/images/training_0_3_640_640_640_640/labels/" --val_images="/data/617/images/training_32_49_640_640_640_640/images/" --val_annotations="/data/617/images/training_32_49_640_640_640_640/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000 --selective_loss=100 --load_weights=1 --start_id=0 --end_id=14

1000 @ 4__non_aug/640_selective

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_train.py --save_weights_path=weights/vgg_unet2_0_3_640_640_640_640_1000 --train_images="/data/617/images/training_0_3_640_640_640_640/images/" --train_annotations="/data/617/images/training_0_3_640_640_640_640/labels/" --val_images="/data/617/images/training_32_49_640_640_640_640/images/" --val_annotations="/data/617/images/training_32_49_640_640_640_640/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000 --selective_loss=1000 --load_weights=1

1000__0_14 @ 4__non_aug/640_selective

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_train.py --save_weights_path=weights/vgg_unet2_0_3_640_640_640_640_0_14_1000 --train_images="/data/617/images/training_0_3_640_640_640_640/images/" --train_annotations="/data/617/images/training_0_3_640_640_640_640/labels/" --val_images="/data/617/images/training_32_49_640_640_640_640/images/" --val_annotations="/data/617/images/training_32_49_640_640_640_640/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000 --selective_loss=1000 --load_weights=1 --start_id=0 --end_id=14

4 @ 640_selective

2 @ 4/640_selective

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_train.py --save_weights_path=weights/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip_2 --train_images="/data/617/images/training_0_3_640_640_64_256_rot_15_345_4_flip/images/" --train_annotations="/data/617/images/training_0_3_640_640_64_256_rot_15_345_4_flip/labels/" --val_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --val_annotations="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000 --selective_loss=2 --load_weights=1

5 @ 4/640_selective

CUDA_VISIBLE_DEVICES=0 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_train.py --save_weights_path=weights/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip_5 --train_images="/data/617/images/training_0_3_640_640_64_256_rot_15_345_4_flip/images/" --train_annotations="/data/617/images/training_0_3_640_640_64_256_rot_15_345_4_flip/labels/" --val_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --val_annotations="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000 --selective_loss=5 --load_weights=1

10 @ 4/640_selective

CUDA_VISIBLE_DEVICES=0 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_train.py --save_weights_path=weights/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip_10 --train_images="/data/617/images/training_0_3_640_640_64_256_rot_15_345_4_flip/images/" --train_annotations="/data/617/images/training_0_3_640_640_64_256_rot_15_345_4_flip/labels/" --val_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --val_annotations="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000 --selective_loss=10 --load_weights=1

rt @ 10/4/640_selective

CUDA_VISIBLE_DEVICES=0 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_train.py --save_weights_path=weights/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip_10_rt --train_images="/data/617/images/training_0_3_640_640_64_256_rot_15_345_4_flip/images/" --train_annotations="/data/617/images/training_0_3_640_640_64_256_rot_15_345_4_flip/labels/" --val_images="/data/617/images/training_32_49_640_640_640_640/images/" --val_annotations="/data/617/images/training_32_49_640_640_640_640/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000 --selective_loss=10 --load_weights=1

100 @ 4/640_selective

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_train.py --save_weights_path=weights/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip_100 --train_images="/data/617/images/training_0_3_640_640_64_256_rot_15_345_4_flip/images/" --train_annotations="/data/617/images/training_0_3_640_640_64_256_rot_15_345_4_flip/labels/" --val_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --val_annotations="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000 --selective_loss=100 --load_weights=1

vis @ 4/640_selective

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip_100/weights_max_val_acc --test_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --output_path="log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip_100/training_32_49_640_640_64_256_rot_15_345_4_flip_max_val_acc/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2"

python3 ../visDataset.py --images_path=/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images --labels_path=/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels --seg_path=log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip_100/training_32_49_640_640_64_256_rot_15_345_4_flip_max_val_acc/raw --save_path=log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip_100/training_32_49_640_640_64_256_rot_15_345_4_flip_max_val_acc/vis --n_classes=3 --start_id=0 --end_id=-1

no_aug @ 4/640_selective

CUDA_VISIBLE_DEVICES=1 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_predict.py --save_weights_path=weights/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip_100/weights_max_val_acc --test_images="/data/617/images/training_32_49_640_640_640_640/images/" --output_path="log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip_100/training_32_49_640_640_640_640_max_val_acc/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2"

python3 ../visDataset.py --images_path=/data/617/images/training_32_49_640_640_640_640/images --labels_path=/data/617/images/training_32_49_640_640_640_640/labels --seg_path=log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip_100/training_32_49_640_640_640_640_max_val_acc/raw --save_path=log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip_100/training_32_49_640_640_640_640_max_val_acc/vis --n_classes=3 --start_id=0 --end_id=-1

stitched @ no_aug/4/640_selective

python3 ../stitchSubPatchDataset.py src_path=/data/617/images/training_32_49/images img_ext=jpg patch_seq_path=log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip_100/training_32_49_640_640_640_640_max_val_acc/raw stitched_seq_path=log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip_100/training_32_49_max_val_acc/raw patch_height=640 patch_width=640 start_id=0 end_id=-1 show_img=0 stacked=0 method=1 normalize_patches=0 img_ext=png

python3 ../visDataset.py --images_path=/data/617/images/training_32_49/images --labels_path=/data/617/images/training_32_49/labels --seg_path=log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip_100/training_32_49_max_val_acc/raw --save_path=log/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip_100/training_32_49_max_val_acc/vis --n_classes=3 --start_id=0 --end_id=-1

1K @ 4/640_selective

CUDA_VISIBLE_DEVICES=0 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_train.py --save_weights_path=weights/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip_1K --train_images="/data/617/images/training_0_3_640_640_64_256_rot_15_345_4_flip/images/" --train_annotations="/data/617/images/training_0_3_640_640_64_256_rot_15_345_4_flip/labels/" --val_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --val_annotations="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000 --selective_loss=1000 --load_weights=1

5K @ 4/640_selective

CUDA_VISIBLE_DEVICES=0 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python2 img_keras_train.py --save_weights_path=weights/vgg_unet2_0_3_640_640_64_256_rot_15_345_4_flip_5K --train_images="/data/617/images/training_0_3_640_640_64_256_rot_15_345_4_flip/images/" --train_annotations="/data/617/images/training_0_3_640_640_64_256_rot_15_345_4_flip/labels/" --val_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --val_annotations="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000 --selective_loss=5000 --load_weights=1

16 @ 640_selective

100 @ 16/640_selective

CUDA_VISIBLE_DEVICES=0 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_train.py --save_weights_path=weights/vgg_unet2_0_15_640_640_64_256_rot_15_345_4_flip_100 --train_images="/data/617/images/training_0_15_640_640_64_256_rot_15_345_4_flip/images/" --train_annotations="/data/617/images/training_0_15_640_640_64_256_rot_15_345_4_flip/labels/" --val_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --val_annotations="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000 --selective_loss=100

32 @ 640_selective

100 @ 32/640_selective

CUDA_VISIBLE_DEVICES=0 CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.0 CUDNN_PATH="/usr/local/cuda-9.0/lib64/libcudnn.so.7" KERAS_BACKEND=theano THEANO_FLAGS="device=cuda0,floatX=float32,dnn.include_path=/usr/local/cuda-9.0/include,dnn.library_path=/usr/local/cuda-9.0/lib64/" python img_keras_train.py --save_weights_path=weights/vgg_unet2_0_31_640_640_64_256_rot_15_345_4_flip_100 --train_images="/data/617/images/training_0_31_640_640_64_256_rot_15_345_4_flip/images/" --train_annotations="/data/617/images/training_0_31_640_640_64_256_rot_15_345_4_flip/labels/" --val_images="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/images/" --val_annotations="/data/617/images/training_32_49_640_640_64_256_rot_15_345_4_flip/labels/" --n_classes=3 --input_height=640 --input_width=640 --model_name="vgg_unet2" --epochs=1000 --selective_loss=100