StrongReID

Bag of Tricks and A Strong ReID Baseline

Citation

The code is built with reid-strong-baseline. Thanks for their great work.

The codes are expanded on a ReID-baseline , which is open sourced by our co-first author Xingyu Liao.

Get Started

The designed architecture follows this guide PyTorch-Project-Template, you can check each folder's purpose by yourself.

1. cd DIVOTrack/Cross_view_Tracking/StrongReID/

2. Install dependencies:

   • pytorch>=0.4
   • torchvision
   • ignite=0.1.2 (Note: V0.2.0 may result in an error)
   • yacs

3. Prepare dataset

DIVOTrack
    └——————datasets
    |        └——————DIVO
    |           |——————ReID_format
    └——————${ROOT}

5. Prepare pretrained model Put the model in ./models/. You can obtain the model from Google Drive

Train

python3 tools/train.py --config_file='configs/softmax_triplet_with_center.yml' MODEL.DEVICE_ID "('your device id')" DATASETS.NAMES "('ReID_format')" OUTPUT_DIR "('your path to save checkpoints and logs')"

Test

Download our final model and put it into ./models.

python3 tools/test.py --config_file='configs/softmax_triplet_with_center.yml' TEST.NECK_FEAT "('after')" TEST.FEAT_NORM "('yes')" MODEL.PRETRAIN_CHOICE "('self')" TEST.RE_RANKING "('yes')" TEST.WEIGHT "('your train model path')"

The test will generate rsb_divo.npy to DIVOTrack/Cross_view_Tracking/StrongReID. If you want to change its name, modify in tools/test.py
Format of rsb_divo.npy

{
circleRegion:{
    Drone:[[fid,pid,lx,ly,w,h,1,0,0,0,feature],...],   
    View1:[...],   
    View2:[...]
}, 
 innerShop:{
    Drone:[[fid,pid,lx,ly,w,h,1,0,0,0,feature],...],   
    View1:[...],   
    View2:[...]
}, 
 ...
 }

Evaluation

Please refer to Multi_view_Tracking