SVLL-ReID

KO | EN

gitlite — search

#javascript #python #hacktoberfest #react #ai #typescript #llm #go #golang #android #machine-learning #rust #deep-learning #linux

★ 14 Open GitHub ↗

No description available.

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Image re-identification: Where self-supervision meets vision-language learning [PDF](https://www.sciencedirect.com/science/article/abs/pii/S0262885625000034)
This is an implementation of Image Re-Identification: Where Self-supervisionMeets Vision-Language Learning  
### Pipeline
<img width="1373" height="750" alt="image" src="https://github.com/user-attachments/assets/dee677d5-39d9-40a1-a330-69458c05235c" />

### Requirements
```
conda create -n svll-reid python=3.8
conda activate svll-reid
conda install pytorch==1.8.0 torchvision==0.9.0 torchaudio==0.8.0 cudatoolkit=10.2 -c pytorch
pip install yacs
pip install timm
pip install scikit-image
pip install tqdm
pip install ftfy
pip install regex
```

### Train
```
    For example,you attempt to train your model on DukeMTMC:

1, modify here: (in configs/person/vit_svllreid.yml)
            ###
            DATASETS:
               NAMES: ('dukemtmc')
            #   NAMES: ('occ_duke')
            #   NAMES: ('market1501')
            #   NAMES: ('msmt17')
               ROOT_DIR: ('/your_dataset_dir')
            OUTPUT_DIR: '/your_output_dir'
            ###

2, and run:
            ###
            CUDA_VISIBLE_DEVICES=0 python train_svlleid.py --config_file configs/person/vit_svllreid.yml
            ###
```
### Test
    CUDA_VISIBLE_DEVICES=0 python test_svllreid.py --config_file configs/person/vit_svllreid.yml TEST.WEIGHT 'your_output_dir/your.pth'

### Important Parameters for SVLL-ReID:
```
Taking the person reid task as an example, you can find svllreid.yml in SAVLL-ReID\Codes\configs\person\config.
Below is the explanation of the key self-supervised parameters for SVLL-ReID:

Language Self-Supervision: STAGE1

Language Self-Supervision Weight: SSL_LOSS_WEIGHT

Method for Augmenting Prompts: PROMPT_AUG_METHOD

Augmentation Strength Rate for Prompts: PROMPT_AUG_SCALE

Vision Self-Supervision: STAGE2

Vision Self-Supervision Weight: SSL_LOSS_WEIGHT

Random Erasing Strength: INPUT: RE_PROB
```

### Enabling and Disabling Language and Vison Self-Supervision:

```
In SVLL-ReID/processor/processor_svllreid_stage1.py, set USE_SSL_stage1 to True or False.

In SVLL-ReID/processor/processor_svllreid_stage2.py, set USE_SSL_stage1 to True or False.
```

### Pre-trained model

### Citation
If you use this code for your research, please cite
```
@article{WANG2025105415,
title = {Image re-identification: Where self-supervision meets vision-language learning},
journal = {Image and Vision Computing},
volume = {154},
pages = {105415},
year = {2025}
}
```

SVLL-ReID

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

SVLL-ReID

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README