// ⬡ Deep Learning Foundations
U-Net
A U-shaped conv net: downsample, upsample, and skip-connect matching levels.
2015 Foundational core
IN PLAIN WORDS
The left half shrinks the image like a normal CNN to capture meaning; the right half upsamples back to full size, and skip connections concatenate same-level features so the output keeps both semantics and detail. Built for biomedical segmentation, it later became the denoiser backbone of DDPM and Stable Diffusion.
Used in: Medical image segmentation; the denoising network in DDPM and Stable Diffusion 1.x/2.x
01Original paper
U-Net: Convolutional Networks for Biomedical Image Segmentation
Olaf Ronneberger, Philipp Fischer, Thomas Brox · 2015
arXiv ↗ ✓ Verified · arXiv API
02Where it sits
Learn the left column first; the right column builds on this one.
Learn first
Then you can learn
- Nothing in the atlas builds on it yet
Related:Denoising Diffusion Probabilistic Models (DDPM)Latent Diffusion ModelsSegment Anything (SAM)
Left: prerequisites · right: builds on this · dashed: related. Full map →
03Learning resources
Official docs, explainers, courses and reference code. All links checked.
04Where the course teaches it
For AI / data
- Verification: verified, checked 2026-10-22.
- JSON: /api/learn/algorithms.json (id:
unet)