Code release for "Cloth-Changing Person Re-identification from A Single Image with Gait Prediction and Regularization". More files are coming soon.

Sorry for delay. Due to an unexpected "Reject" received from reviewer, here we only provide the metarials for Baseline now (including dataloader files for multiple colth-changing reid datasets). The rest parts about gait prediction and regularization will be released upon acceptance.

nohup python train_imgreid_xent_htri.py --gpu-devices 1 --root /data1/Datasets_jinx/Cloth-ReID/reid_datasets/ -s ltcc -t ltcc -a resnet50_fc512 --height 256 --width 128 --max-epoch 60 --stepsize 20 40 --lr 0.0003 --train-batch-size 32 --test-batch-size 100 --save-dir ../debug_dir/test_ltcc3 --train-with-all-cloth --use-standard-metric --eval-freq 10 > ./debug3.log 2>&1 &

nohup python train_imgreid_xent_htri.py --gpu-devices 1 --root /data1/Datasets_jinx/Cloth-ReID/reid_datasets/ -s ltcc -t ltcc -a resnet50_fc512 --height 256 --width 128 --max-epoch 60 --stepsize 20 40 --lr 0.0003 --train-batch-size 32 --test-batch-size 100 --save-dir ../debug_dir/test_ltcc3 --train-with-all-cloth --use-cloth-changing-metric --eval-freq 10 > ./debug5.log 2>&1 &

nohup python train_imgreid_xent_htri.py --gpu-devices 1 --root /data1/Datasets_jinx/Cloth-ReID/reid_datasets/ -s prcc -t prcc -a resnet50_fc512 --height 256 --width 128 --max-epoch 60 --stepsize 20 40 --lr 0.0003 --train-batch-size 32 --test-batch-size 100 --save-dir ../debug_dir/test_prcc --same-clothes --just-for-prcc-test --use-metric-cuhk03 --eval-freq 10 > ./debug8_prcc.log 2>&1 &

nohup python train_imgreid_xent_htri.py --gpu-devices 1 --root /data1/Datasets_jinx/Cloth-ReID/reid_datasets/ -s prcc -t prcc -a resnet50_fc512 --height 256 --width 128 --max-epoch 60 --stepsize 20 40 --lr 0.0003 --train-batch-size 32 --test-batch-size 100 --save-dir ../debug_dir/test_prcc --cross-clothes --just-for-prcc-test --use-metric-cuhk03 --eval-freq 10 > ./debug9_prcc.log 2>&1 &