Computer Vision Section 060
Segmentation in Depth
Labelling every single pixel: the three kinds of segmentation, the architectures behind each, and the messy masks they return.
11 of 11 lessons published Three reading levels on every lesson
Lessons in order
Work top to bottom. Each lesson assumes the one above it.
- Semantic, instance and panoptic segmentation
- U-Net
- DeepLab and atrous convolution
- Mask R-CNN
- Panoptic segmentation
- Mask2Former and universal segmentation
- Segment Anything
- Dice and other segmentation losses
- Weakly supervised segmentation
- Cleaning up predicted masks
- Image matting and background removal