Fake Image Detector
Image Tampering Detection using ELA and CNN
Members
Indonesian paper/documentation
Project objective
Combine the implementation of error-level analysis (ELA) and deep learning to detect whether an image has undergone fabrication or/and editing process or not, e.g. splicing.
Methods
- Error-level analysis
- Convolutional neural networks (CNN)
Architecture
Result
- Convergence: Epoch 9
- Best accuracy: 91.83% (epoch 9)
Dataset
Please refer to issue #1.