Strong Person ReID Baseline

The architecture follows the NTU ROSE ReID Project guide line Person_ReID_Baseline. Some of codes are copy from L1aoXingyu's reid_baseline.

• ResNet50 Last Stride 1 from huanghoujing's triplet baseline
• WarmupMultiStepLR: is from FAIR's paper: 'Accurate, Large Minibatch SGD: Training ImageNet in 1 Hour'

Requirements

• python 3
• pytorch 1.0 + torchvision
• yacs Yet Another Configuration System
• fire Automatically generating command line interfaces (CLIs)

Install all dependences libraries

pip3 install -r requirements.txt

Configs

Use different yaml config files for different experiment settings. All the config files are store in folder config. Please use different OUTPUT_DIR names for different experiments to avoid conflit and accidentally files overwritten.

Datasets

This code support CUHK03, Market1501, DukeMTMC and MSMT17 datasets. All these dataset should be defined in the DATASETS.NAMES of the config file, our code will be download the corresponding dataset automatically (into the datasets folder). As this fuction require access to Google Drive, it will not work in China. Currently support:

• CUHK03
• Market1501
• DukeMTMC
• MSMT17

Training:

• As the last feature maps increase. Batch Size 128 will use more than 12G
• Batch Size 64 use around 10G GPU memory

python train.py ./config/market_softmax.yaml

### Change GPU
python train.py ./config/market_softmax.yaml --DEVICE=cuda:5

Testing:

### No Re-Ranking
python test.py ./config/market_softmax.yaml

### Change GPU
python test.py ./config/market_softmax.yaml --DEVICE=cuda:5

### With Re-Ranking
python test.py ./config/market_softmax.yaml --RE_RANKING=True

Testing Cross Dataset:

### Market1501 -> DukeMTMC
python test_cross_dataset.py ./config/market_softmax.yaml DukeMTMC

Results Compare with Person ReID Baseline

Softmax Only Batch Size 64: Rank1 (mAP)

Dataset	Softmax	Strong Softmax
CUHK03	56.1 (52.4)	61.1 (56.2
Market1501	91.6 (78.7)	92.5 (80.2)
DukeMTMC	83.4 (66.6)	84.8 (68.3)
MSMT17	69.0 (40.1)	71.4 (42.5)

Softmax+Tripelt Only Batch Size 64 : Rank1 (mAP)

	Softmax+Triplet	Strong Softmax+Triplet
CUHK03	65.6 (61.8)	66.3 (61.8)
Market1501	93.2 (82.0)	93.4 (83.1)
DukeMTMC	86.4 (72.4)	86.2 (72.5)
MSMT17	73.9 (46.4)	74.6 (47.3)