PyTorchVideo: A deep learning library for video understanding
AI News Sharing
PyTorchVideo is a deep learning library for research and applications in video understanding.It provides reusable, modular and efficient components needed to accelerate the video understanding research.
Below image is the example of a PyTorchVideo-based SlowFast model performing video action detection.
Image source: https://github.com/facebookresearch/pytorchvideo
Pytorch also provides comprehensive benchmarks to evaluate the supported models on different datasets using standard evaluation setup. All the models can be downloaded from the provided links.
Based on the example PyTorch provided, you can use model builder to build selected model and deploy model easily. You can also modify the dataset to suit your need and build your own training model.
While training your new model, it would consume lots GPU memory and CUDA resources running massive 3D object datasets.
Get your hardware ready and explore more on PyTorchVideo!