A Review of Keypoints’ Detection and Feature Description in Image Registration

Figure 3

SuperPoint joint training. Train a fully convolutional network that jointly extracts interest points and descriptors from an image.