Classify actions in video with TAO.
video action recognition
Trains, evaluates, exports, and runs inference on video action-recognition models using RGB, optical-flow, or joint inputs.
When to use it
Use for training, evaluating, exporting, or running inference on a TAO action-recognition model.
Give it the required datasets, specifications, and checkpoints; it runs the requested action and writes results under the working directory.
What you provide
This skill
TAO sample archives
Downloads TAO sample archives
Docker is required to launch the TAO container.
nvidia-container-toolkit is required for GPU-enabled container execution.
Requires at least one GPU with 16GB or more VRAM per GPU; two GPUs are recommended.
Standalone sessions not initialized by the TAO skill bank plugin must run tao-setup first.