SOTA Re-identification Methods and Toolbox

apex baseline computer-vision image-retrieval image-search open-reid person-reid person-reidentification pytorch random-erasing re-identification re-ranking reids sota toolbox

Go to file

liaoxingyu 3a22aa52f3 update		2018-07-26 08:18:25 +08:00
bases	update	2018-07-21 17:19:33 +08:00
datasets	first update	2018-06-08 12:59:03 +08:00
models	update	2018-07-21 17:19:33 +08:00
scripts	update	2018-07-21 17:19:33 +08:00
trainers	update	2018-07-21 17:19:33 +08:00
utils	first update	2018-06-08 12:59:03 +08:00
.gitignore	update readme	2018-06-15 17:49:47 +08:00
README.md	update	2018-07-26 08:18:25 +08:00
config.py	update	2018-07-21 17:19:33 +08:00
main_reid.py	update	2018-07-26 08:18:25 +08:00

README.md

ReID_baseline

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

Learning rate

This is the warmpup strategy learning rate

Results

loss	rank1	map
softmax	87.9%	70.1%
triplet	88.8%	74.8%
triplet + softmax	92.0%	78.1%