Re-ID

Market 1501

Resnet50 : batchsize=32 : Rank1 =95.81 , mAP=88.28

If you want a higher accuracy, you may modify batchsize or use Resnet101.

DukeMTMC-REID

Resnet50: batchsize =32 : Rank1 =88.7 , mAP=77.9

pytorch = 0.4

Name		Name	Last commit message	Last commit date
Latest commit History 31 Commits
README.md		README.md
demo.py		demo.py
evaluate_gpu.py		evaluate_gpu.py
evaluate_rerank.py		evaluate_rerank.py
lr_scheduler.py		lr_scheduler.py
model.py		model.py
prepare.py		prepare.py
random_erasing.py		random_erasing.py
re_ranking.py		re_ranking.py
samplers.py		samplers.py
test.py		test.py
train_11.py		train_11.py
triplet_loss.py		triplet_loss.py

Provide feedback