spmallick/learnopencv

Covers real-world computer vision and AI applications including object detection (YOLO variants), multi-object tracking, face detection/blur, 3D reconstruction (SAM-3, Gaussian splatting), and vision-language models (VLMs) deployed on edge devices like Jetson. Code examples span detection pipelines, RAG systems for video understanding, fine-tuning workflows, and LLM inference optimization, with implementations in both C++ and Python using OpenCV's DNN module alongside frameworks like PyTorch and vLLM.

22,783 stars. Actively maintained with 35 commits in the last 30 days.