Skip to content

Latest commit

 

History

History
42 lines (26 loc) · 1.25 KB

README.md

File metadata and controls

42 lines (26 loc) · 1.25 KB

Fake-Image-Detection

*Disclaimer:- Dataset and Neural Network Model not provided in this repo.

Application Front:-

Application Front

Application Results Page:-

Application Results

This project uses python libraries like Keras and Tensorflow to tell If an Image has been manipulated or not. I chose the approach of using transfer learning for getting more accurate results.In this project the images are firstly connverted into Error Level Analysis(ELA) images and then fed to the neural network for training.I used the VGG16 model for transfer learning.

This project requires the following dependencies:-

  • EEl - Used for displaying webpages using a local host environment,it also allows python functions to be implemented in JavaScript
  • Keras
  • Tensorflow
  • Pillow
  • Numpy
  • Pandas
  • Tkinter

Results:-

Training Graph:-

Training Graph

Validation Results:-

Validation

Testing Results:-

Testing

Futher Things That can be Enhanced:-

  • Improving the accuracy of the neural network
  • Increasing the dataset and then training the neural network.
  • Using a different model like ResNet50 for transfer learning.