Semantic segmentation

Level 5: Segmentation and dense prediction

Coming soon

This page is being written. See the full roadmap for the whole syllabus.

What this chapter will cover

• Per-pixel classification
• fully convolutional networks
• encoder-decoder and U-Net
• skip connections
• transposed convolution and upsampling
• dilated convolution and DeepLab
• losses (cross-entropy, Dice, focal)
• metrics (pixel accuracy, IoU, Dice)