Technical Resources

Yuanjun Xiong

Last update: September, 2016

Technical Reports

• CES-START@ActivityNet 2016
  • Champion team of ActivityNet large scale action recognition challenge.
• DeepID-Net
  • Winning team for ILSVRC14
• Towards Good Practices for Very Deep Two-Stream ConvNets
  • Current state-of-the-art for video-based action recognition. Code and models available. Project Site

Articles

• Compile and Debug CUDA enabled mex-file using Visual Studio
• Install and Run cuda-convnet on Windows OS with VS2010

Code

• Temporal Segment Networks
  • A full-fledged action recognition framework based on state-of-the-art action recognition technique: temporal segment networks.
• A Caffe fork with parallel training of multiple GPUs enabled

Return to home page