Awesome

MSINet

Official implementation of "MSINet: Twins Contrastive Search of Multi-Scale Interaction for Object ReID".

Highlight

• MSINet is accepted by CVPR2023.
• MSINet is a light-weighted network architecture for object Re-ID tasks.
• MSINet trained from scratch achieves higher retrieval performance compared with pre-trained ResNet-50.

Abstract

Neural Architecture Search (NAS) has been increasingly appealing to the society of object Re-Identification (ReID), for that task-specific architectures significantly improve the retrieval performance. Previous works explore new optimizing targets and search spaces for NAS ReID, yet they neglect the difference of training schemes between image classification and ReID. In this work, we propose a novel Twins Contrastive Mechanism (TCM) to provide more appropriate supervision for ReID architecture search. TCM reduces the category overlaps between the training and validation data, and assists NAS in simulating real-world ReID training schemes. We then design a Multi-Scale Interaction (MSI) search space to search for rational interaction operations between multi-scale features. In addition, we introduce a Spatial Alignment Module (SAM) to further enhance the attention consistency confronted with images from different sources. Under the proposed NAS scheme, a specific architecture is automatically searched, named as MSINet. Extensive experiments demonstrate that our method surpasses state-of-the-art ReID methods on both in-domain and cross-domain scenarios.

Datasets

Put the datasets into ./data

• Market-1501
• MSMT17
• VeRi-776
• VehicleID

Experiment Commands

Evaluate the Performance of MSINet

Train from Scratch

python train.py

Train after Pre-training

Download the pre-trained model from Google Drive and put it into ./pretrained.

python train.py --pretrained

Train Cross-domain Experiments

python train.py -ds market1501 -dt msmt17 --pretrained --epochs 250

Add SAM for Cross-domain Experiments

python train.py -ds market1501 -dt msmt17 --pretrained --epochs 250 --sam-mode pos_neg

To train vehicle datasets, please add --width 256.

Conduct Search for Other Re-ID Datasets

python search.py -ds market1501

Citation

@inproceedings{gu2023msinet,
  title={MSINet: Twins Contrastive Search of Multi-Scale Interaction for Object ReID},
  author={Gu, Jianyang and Wang, Kai and Luo, Hao and Chen, Chen and Jiang, Wei and Fang, Yuqiang and Zhang, Shanghang and You, Yang and Zhao, Jian},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={19243--19253},
  year={2023}
}