// the find
vipstone/faceai
一款入门级的人脸、视频、文字检测以及识别的项目.
A beginner-oriented computer vision demo project covering face detection, recognition, emotion classification, gender detection, image colorization, and basic OCR. It's a collection of standalone Python scripts that each demonstrate one technique using whatever library was popular in 2018. Aimed at Chinese-speaking developers learning CV fundamentals.
Each feature is a self-contained script — easy to read and run independently without understanding the whole project. The included pre-trained models (emotion CNN at 65% accuracy, gender XCEPTION at 95%) mean you can run demos without any training setup. Documentation is thorough for its era, with step-by-step setup guides for Windows, Dlib, Tesseract, and OpenCV. Covers an unusually wide range of CV tasks in one place, which is useful as a reference map of what's possible.
Pinned to Python 3.6 and TensorFlow 1.8 — both end-of-life years ago. Getting this running on a modern machine means fighting dependency hell across Dlib, Keras 2.1, and TF 1.x, which is a significant time sink with no guarantee of success. Last commit was April 2020; face swap and eye tracking are explicitly marked unfinished and were never completed. No requirements.txt or environment file — the README lists library versions but you're on your own to reconstruct the environment. The scripts are demos, not a library; there's nothing to import or build on.