
sherlockchou86/VideoPipe
🏗️ Frameworksherlockchou86
A cross-platform video analysis framework for object detection, face recognition, and LLM integration.
VideoPipe is a cross-platform video analysis framework written in C++. It provides a modular pipeline architecture built on GStreamer and OpenCV, enabling developers to construct complex video processing flows. Key features include multi-model support for tasks such as object detection (YOLO, etc.), face recognition, license plate recognition, image segmentation, classification, ReID, and similarity search. The framework integrates with DeepStream for GPU acceleration and supports LLM backends like Ollama and OpenAI for multimodal analysis. VideoPipe abstracts common video analysis patterns, allowing users to focus on business logic rather than low-level stream handling. The project is open-source and actively maintained, with extensive documentation and examples.
💡Highlights
- ├─Cross-platform video analysis framework
- ├─Supports object detection, face recognition, license plate
- └─Integrates with LLMs (Ollama, OpenAI)
🎯For
- ├─CV developers
- ├─AI engineers
- └─Video analytics researchers