video-caffe github地址:https://github.com/chuckcho/video-caffe
编译过程:
Key steps to build video-caffe are:
- 1.git clone https://github.com/chuckcho/video-caffe.git
- 2.cd video-caffe
- 3.cp Makefile.config.example Makefile.config
- 4.Make sure CUDA and CuDNN are detected and their paths are correct.(去掉 use cudnn :=1前面的#号)
- 5.make all -j8(线程数)
编译完成。
UCF-101 training demo
Scripts and training files for C3D training on UCF-101 are located in examples/c3d_ucf101/. Steps to train C3D on UCF-101:
- 1.Download UCF-101 dataset from UCF-101 website.
- 2.Unzip the dataset: e.g. unrar x UCF101.rar
- 3.(Optional) video reader works more stably with extracted frames than directly with video files. Extract frames from UCF-101 videos by revising and running a helper script,{video-caffe-root}/examples/c3d_ucf101/extract_UCF-101_frames.sh
(也可用ffmpeg命令) - 4.Change ${video-caffe-root}/examples/c3d_ucf101/c3d_ucf101_{train,test}_split1.txt to correctly point to UCF-101 videos or directories that contain extracted frames.
- 5.Modify ${video-caffe-root}/examples/c3d_ucf101/c3d_ucf101_train_test.prototxt to your taste or HW specification. Especially batch_size may need to be adjusted for the GPU memory.
- 6.Run training script: e.g. cd ${video-caffe-root} && examples/c3d_ucf101/train_ucf101.sh (optionally use –gpu to use multiple GPU’s)
- 7.(Optional) Occasionally run ${video-caffe-root}/tool/extra/plot_training_loss.sh to get training loss / validation accuracy (top1/5) plot. It’s pretty hacky, so look at the file to meet your need.(caffe/tools/extra/plot_training_log.py.example自带绘图的小脚本)
- 8.At 7 epochs of training, clip accuracy should be around 45%.