Unsupervised image segmentation — approaches.

hard

Answer

$(1) Classical: k-means or mean-shift on pixel color + position (SLIC superpixels).$
$(2) Spectral clustering on affinity graph.$
$(3) Modern DL: DINO's self-attention maps segment objects without any supervision.$
$(4) SAM (Segment Anything Model, Meta 2023): promptable segmentation trained on 1B masks — used with 'automatic mask' mode for full-image unsupervised segmentation.$
$(5) STEGO (Hamilton et al. 2022): contrastive semantic segmentation.$

Check yourself — multiple choice

Random
k-means/SLIC on pixels + position / spectral / DINO self-attention (emergent!) / SAM automatic-mask mode / STEGO contrastive semantic seg
Only supervised
Not real

Unsup segmentation: SLIC / spectral / DINO / SAM automatic / STEGO.

#applications

Practise Unsupervised Learning

214 interview questions in this topic.

Read the lesson Flashcards Take the quiz

Related questions