Download presentation
Presentation is loading. Please wait.
1
Action Recognition
2
Dataset UCF101 HMDB51 Kinetics
3
HMDB51 51 classes 7,000 clips
4
Kinetics 400 classes 300,000 clips
5
Architectures 3D Convnet 2D convnet → LSTM
6
3D Convnet Uses 3d kernel to interpret temporal data
Is slower to train as it
7
2D Convnet → LSTM Can use image recongnition 2D convnet as a starting point to speed training Can be very deep due to using LSTM
8
Python, Tensorflow, Caffe, Examples
Comfortable with Python Have used Tensorflow Need to finish installing Caffe Coding examples. ( HMDB51
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.