Home

Awesome

ReID_baseline

Baseline model (with bottleneck) for person ReID (using softmax and triplet loss).

We support

Get Started

  1. cd to folder where you want to download this repo

  2. Run git clone https://github.com/L1aoXingyu/reid_baseline_gluon.git

  3. Install dependencies:

    pip install --pre mxnet-cu90
    
    • tensorflow (for tensorboard)
    • MXBoard
  4. Prepare dataset

    Create a directory to store reid datasets under this repo via

    cd reid_baseline
    mkdir data
    
    1. Download dataset to data/ from http://www.liangzheng.org/Project/project_reid.html
    2. Extract dataset and rename to market1501. The data structure would like:
    market1501/
        bounding_box_test/
        bounding_box_train/
    
  5. Prepare pretrained model if you don't have

    from mxnet import gluon
    gluon.model_zoo.vision.resnet50_v1(pretrained=True)
    

    Then it will automatically download model in ~.mxnet/models/, you should set this path in config.py

Train

You can run

bash scripts/train_triplet_softmax.sh

in reid_baseline folder if you want to train with softmax and triplet loss. You can find others train scripts in scripts.

Results

lossrank1map
softmax87.1%67.8%
triplet88.2%73.7%
triplet + softmax90.4%76.4%

I find the mxnet.gluon results are a little bit lower than pytorch results, and I cannot get the reason. I would appreciate that if anyone can help me.