Added few function

SoloSynthGAN/functions.py

@@ -17,6 +17,51 @@
 
 from SoloSynthGAN.resize_image import imresize, imresize_in, imresize_to_shape
 
+
+def image_2_video(dir2save, netG, fixed_noise, reals, noise_amp, opt, alpha=0.1, beta=0.9, start_scale=1,
+                 num_images=100, fps=10):
+
+    def denorm_for_gif(img):
+        img = denorm(img).detach()
+        img = img[0, :, :, :].cpu().numpy()
+        img = img.transpose(1, 2, 0) * 255
+        img = img.astype(np.uint8)
+        return img
+
+    reals_shapes = [r.shape for r in reals]
+    all_images = []
+
+    with torch.no_grad():
+        noise_random = sample_random_noise(len(fixed_noise) - 1, reals_shapes, opt)
+
+
+        z_prev1 = [0.99 * fixed_noise[i] + 0.01 * noise_random[i] for i in range(len(fixed_noise))]
+
+
+        z_prev2 = fixed_noise
+
+        for _ in range(num_images):
+
+            noise_random = sample_random_noise(len(fixed_noise)-1, reals_shapes, opt)
+            diff_curr = [beta*(z_prev1[i]-z_prev2[i])+(1-beta)*noise_random[i] for i in range(len(fixed_noise))]
+            z_curr = [alpha * fixed_noise[i] + (1 - alpha) * (z_prev1[i] + diff_curr[i]) for i in range(len(fixed_noise))]
+
+            if start_scale > 0:
+                z_curr = [fixed_noise[i] for i in range(start_scale)] + [z_curr[i] for i in range(start_scale, len(fixed_noise))]
+
+            z_prev2 = z_prev1
+            z_prev1 = z_curr
+
+            sample = netG(z_curr, reals_shapes, noise_amp)
+
+            sample = denorm_for_gif(sample)
+
+            all_images.append(sample)
+
+
+    imageio.mimsave('{}/start_scale={}_alpha={}_beta={}.gif'.format(dir2save, start_scale, alpha, beta), all_images, duration=int(1000*(1/fps)))
+
+
 def denorm(x):
     out = (x + 1) / 2
     return out.clamp(0, 1)
@@ -57,7 +102,7 @@ def generate_noise(size,num_samp=1,device='cuda',type='gaussian', scale=1):
 
 
 def upsampling(im,sx,sy):
-    m = nn.Upsample(size=[round(sx),round(sy)],mode='bilinear',align_corners=True)
+    m = nn.Upsample(size=[round(sx),round(sy)],mode='bicubic',align_corners=True)
     return m(im)
 
 

main.py

@@ -33,7 +33,7 @@ def get_arguments():
     parser.add_argument('--start_scale', type=int, help='at which stage to start training', default=0)
 
 
-    parser.add_argument('--niter', type=int, default=800, help='number of epochs to train per scale')
+    parser.add_argument('--niter', type=int, default=1500, help='number of epochs to train per scale')
     parser.add_argument('--gamma',type=float,help='scheduler gamma',default=0.1)
     parser.add_argument('--lr_g', type=float, default=0.0005, help='learning rate, default=0.0005')
     parser.add_argument('--lr_d', type=float, default=0.0005, help='learning rate, default=0.0005')
@@ -59,14 +59,16 @@ def get_arguments():
     parser.add_argument('--train_mode', default='generation',
                         choices=['generation', 'retarget', 'harmonization', 'editing', 'animation'],
                         help="generation, retarget, harmonization, editing, animation")
-    parser.add_argument('--lr_scale', type=float, help='scaling of learning rate for lower stages', default=0.12)
-    parser.add_argument('--train_stages', type=int, help='how many stages to use for training', default=9)
+    parser.add_argument('--lr_scale', type=float, help='scaling of learning rate for lower stages', default=0.5)
+    parser.add_argument('--train_stages', type=int, help='how many stages to use for training', default=6)
 
     parser.add_argument('--fine_tune', action='store_true', help='whether to fine tune on a given image', default=0)
     parser.add_argument('--model_dir', help='model to be used for fine tuning (harmonization or editing)', default="")
 
     opt = parser.parse_args()
+    print(opt)
     opt = functions.post_config(opt)
+
 
     if opt.fine_tune:
         _gpu = opt.gpu