
RobustVideoMatting
🧠 AI ModelPeterL1n
Robust video matting in multiple frameworks: PyTorch, TensorFlow, ONNX, CoreML.
RobustVideoMatting (RVM) is a neural network architecture designed for real-time video matting. It uses a recurrent architecture with a temporal feature aggregation module to maintain temporal coherence across frames. The model is trained on a large dataset with diverse scenes. Key features include high-resolution output (up to 4K), support for mobile devices via CoreML and TensorFlow.js, and integration with popular frameworks. It achieves state-of-the-art performance on benchmarks like Distinction-646 and has been used in production.
💡Highlights
- ├─PyTorch, TensorFlow, ONNX, CoreML
- ├─Real-time 4K video matting
- └─Temporal consistency across frames
🎯For
- ├─Computer vision researchers
- ├─Video editors
- └─Mobile app developers