deep-learning-for-image-processing

Why we included this project

This is a Chinese-language tutorial series that grew out of the author's graduate research, and it covers the standard computer vision stack in a way few single repos do. Each architecture gets a Bilibili lecture explaining the structure and the ideas behind it, plus working code, and the code is written twice: once in PyTorch and once in TensorFlow 2. So a learner can go from LeNet and ResNet through Faster R-CNN and the YOLO detectors to segmentation models like FCN, DeepLabV3, U-Net, and Mask R-CNN, and see the same network built in both major frameworks. The slide decks and videos make it feel like a real course rather than a pile of notebooks, which is handy both as a primary study path and as a place to look up how a specific architecture is wired. Teams onboarding new people into vision work will find the step-by-step walkthroughs a fast way to get everyone productive.