Self-Supervised Learning


Introduced by Pathak et al. in Context Encoders: Feature Learning by Inpainting

Train a convolutional neural network to generate the contents of an arbitrary image region conditioned on its surroundings.

Source: Context Encoders: Feature Learning by Inpainting


Paper Code Results Date Stars


Task Papers Share
Image Inpainting 198 23.08%
Image Generation 54 6.29%
Denoising 48 5.59%
Video Inpainting 41 4.78%
Semantic Segmentation 24 2.80%
Image Restoration 20 2.33%
Facial Inpainting 20 2.33%
Super-Resolution 19 2.21%
Optical Flow Estimation 19 2.21%


Component Type
🤖 No Components Found You can add them if they exist; e.g. Mask R-CNN uses RoIAlign