GitHub - jason718/RodentBehaviorAnnotation: Code For Ren et al. WACV 2017 (original) (raw)

Who moved my cheese? Automatic Annotation of Rodent Behaviors with Convolutional Neural Networks

[Project] [Paper]

If you use this work, please cite:

@inproceedings{ren-wacv2017, title = {Who moved my cheese? Automatic Annotation of Rodent Behaviors with Convolutional Neural Networks}, author = {Ren, Zhongzheng and Noronha, Adriana and Ciernia, Annie Vogel and Lee, Yong Jae}, booktitle = {Winter Conference on Applications of Computer Vision (WACV)}, year = {2017} }

Setup

Prerequisites

• Linux
• NVIDIA GPU + CUDA CuDNN Recommended(CPU mode and CUDA without CuDNN also work)

Getting Started

• Install [Caffe]
• Install [C3D]
• Install [pythonVideoAnnotator]
• Clone this repo

Dataset

Please visit this [google drive folder]

Running

Annotating Your videos

You can use the UI in this code to annotate your own videos. It can be used to annotate any videos with dense frame-level action labels. We also provide the tools to parse the annotations and generate txt files for caffe to use.

Training

We fine-tune an alexnet and C3d model here. Basically, change the softmax to 5 class and fine-tune the whole networks.

Testing

Contact

If you meet any trouble when using this program, fell free to contact me!