einiBasketball

It is for the code and dataset for the Re-Identification Algorithm for Multiplayer Players Based on Sliding Window and Weighting-Scoring.

env setting

conda create -n sw2mp python=3.7
conda activate sw2mp
conda install pytorch==1.1.0 torchvision==0.3.0 -c pytorch

GPU Memmory >= 10G, Memory >= 20G(LINUX with CUDN)

Pre-trained Model
(pose_hrnet_w48_256x192.pth, r50_ibn_a.pth)
please download it to path r50_ibn_a.pth : core/models/model_reid.py line 71 pose_hrnet_w48_256x192.pth: core/models/model_keypoints/config/default.py line 133)

train

python main.py --mode train \
--dataset_path /your path to daraset \
--output_path ./results

test

python main.py --mode test \
--resume_test_path path/to/pretrained/model --resume_test_epoch 119 \
--duke_path path/to/dataset --output_path ./results

example

python main.py --mode test \
--resume_test_path /root/autodl-tmp/SWWS-Net-master/base/models --resume_test_epoch 119 \
--dataset_path /root/autodl-tmp/dataset/duke --output_path ./exp_on_c

A self-made dataset for basketball players Re-ID. https://github.com/jiangtaoluo/einiBasketball/releases/tag/Datasets