TSN-Pytorch-gesture-recognition

This a gesture recognition experiment in vedio for a logitics project.We use TSN model to discern action from a video in trunk.We can get driver's driving video by camera in the trunk to judge whether the driver has illegal manner when driving.The origin paper can be found here. For more detail about the paper and code, see this [blog][1]

---

setup

• requirements: pytorch1.0, opencv-python, opencv-c++ we use opencv-c++ to extract optical flow

demo

• put your images in data/demo, the results will be saved in data/results, and run demo in the root

python ..//tools/demo.py

---

training

prepare data

• We use own private data to train tsn model, but sorry we couldn't open these data. You can use International public dataset,such as UFC101,hmdb51 e.g.
• Second, you need to download pretrain models adn put it in models.
• We use three input modes to train our models,for RGB mode, optical mode, RGB-Diff mode.

for RGB mode:

python main.py ucf101 RGB <ucf101_rgb_train_list> <ucf101_rgb_val_list> \
   --arch BNInception --num_segments 3 \
   --gd 20 --lr 0.001 --lr_steps 30 60 --epochs 80 \
   -b 128 -j 8 --dropout 0.8 \
   --snapshot_pref ucf101_bninception_ 

For flow models:

python main.py ucf101 Flow <ucf101_flow_train_list> <ucf101_flow_val_list> \
   --arch BNInception --num_segments 3 \
   --gd 20 --lr 0.001 --lr_steps 190 300 --epochs 340 \
   -b 128 -j 8 --dropout 0.7 \
   --snapshot_pref ucf101_bninception_ --flow_pref flow_  

For RGB-diff models:

python main.py ucf101 RGBDiff <ucf101_rgb_train_list> <ucf101_rgb_val_list> \
   --arch BNInception --num_segments 7 \
   --gd 40 --lr 0.001 --lr_steps 80 160 --epochs 180 \
   -b 128 -j 8 --dropout 0.8 \
   --snapshot_pref ucf101_bninception_ 

Testing

After training, there will checkpoints saved by pytorch, for example ucf101_bninception_rgb_checkpoint.pth.

Use the following command to test its performance in the standard TSN testing protocol:

python test_models.py ucf101 RGB <ucf101_rgb_val_list> ucf101_bninception_rgb_checkpoint.pth \
   --arch BNInception --save_scores <score_file_name>

Or for flow models:

python test_models.py ucf101 Flow <ucf101_rgb_val_list> ucf101_bninception_flow_checkpoint.pth \
   --arch BNInception --save_scores <score_file_name> --flow_pref flow_