【行人重识别】TransReID复现（market1501为例）

秒速五厘米 2022-11-15 05:47 790阅读 0赞

代码参考：[https://github.com/heshuting555/TransReID][https_github.com_heshuting555_TransReID]

论文参考：[https://arxiv.org/abs/2102.04378][https_arxiv.org_abs_2102.04378]

# 1.环境 #

ubuntu16.04
    python3.6
    cuda110
    
    torch==1.7.0+cu110
    torchvision==0.8.0
    
    timm
    yacs
    opencv-python==4.1.0.25

# 2.模型准备 #

[https://www.kaggle.com/abhinand05/vit-base-models-pretrained-pytorch][https_www.kaggle.com_abhinand05_vit-base-models-pretrained-pytorch]

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM1OTc1NDQ3_size_16_color_FFFFFF_t_70][]

# 3.修改 #

将下面路径修改为你自己的路径：

1）预训练模型路径

2）数据路径

3）模型保存路径

修改configs/Market/vit\_transreid.yml文件，都改为自己的（最后一个是因为模型保存在TransReID上一级目录中）：

MODEL:  
      PRETRAIN_PATH: '/home/***/.cache/torch/checkpoints/jx_vit_base_p16_224-80ecf9dd.pth'
    
    DATASETS:
      NAMES: ('market1501')
      ROOT_DIR: ('/root/datasets/')
    
    TEST:
      EVAL: True
      IMS_PER_BATCH: 256
      RE_RANKING: False
      WEIGHT: './logs/market_vit_transreid/transformer_120.pth'
    
    OUTPUT_DIR: './logs/market_vit_transreid'

# 4.训练 #

python train.py --config_file configs/Market/vit_transreid.yml MODEL.DEVICE_ID "('0')"

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM1OTc1NDQ3_size_16_color_FFFFFF_t_70 1][]

![20210331130141674.png][]

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM1OTc1NDQ3_size_16_color_FFFFFF_t_70 2][]

[https_github.com_heshuting555_TransReID]: https://github.com/heshuting555/TransReID
[https_arxiv.org_abs_2102.04378]: https://arxiv.org/abs/2102.04378
[https_www.kaggle.com_abhinand05_vit-base-models-pretrained-pytorch]: https://www.kaggle.com/abhinand05/vit-base-models-pretrained-pytorch
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM1OTc1NDQ3_size_16_color_FFFFFF_t_70]: /images/20221022/f3b6ad272f204511b34c948d46e14839.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM1OTc1NDQ3_size_16_color_FFFFFF_t_70 1]: /images/20221022/0ea180203e8440eab7d287010b7dac84.png
[20210331130141674.png]: https://img-blog.csdnimg.cn/20210331130141674.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM1OTc1NDQ3_size_16_color_FFFFFF_t_70 2]: https://img-blog.csdnimg.cn/20210331143844835.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM1OTc1NDQ3,size_16,color_FFFFFF,t_70