CaptainBlackboard

Why we included this project

This is a personal notebook from a computer vision engineer at Alibaba, built from problems that came up in real work rather than textbook exercises. Entries cover depthwise separable convolutions in MobileNet, batch normalization internals, estimating CNN compute and memory, imbalanced training data, attention mechanisms, and knowledge distillation, and each one tends to explain why a technique works, not just what it does. Most of the writing is in Chinese, which makes it a good fit for Chinese-speaking developers who want to firm up their fundamentals or prepare for CV interviews. There is no runnable codebase here, just a long series of notes and code snippets you can read straight through to close gaps in your own understanding.