Sports and human action recognition in videos.
ResNet 3D is a network with 3D convolutions used for video understanding.
Model checkpoint:Kinetics-400
Input resolution:112x112
Number of parameters:33.4M
Model size (float):127 MB
Model size (w8a8):32.1 MB
Camera
Action Recognition
Source Model: BSD-3-CLAUSE
Deployable Model: AI-HUB-MODELS-LICENSE