DC-Former: Diverse and Compact Transformer for Person Re-identification

Authors: Wen Li, Cheng Zou, Meng Wang, Furong Xu, Jianan Zhao, Ruobing Zheng, Yuan Cheng, Wei Chu

AAAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The experimental results of our method are promising, which surpass previous state-of-the-art methods on several commonly used person Re ID benchmarks.
Researcher Affiliation Collaboration 1Ant Group 2Artificial Intelligence Innovation and Incubation (AI3) Institute, Fudan University {yinian.lw,wuyou.zc,darren.wm,booyoungxu.xfr,zhaojianan.zjn,zhengruobing.zrb, weichu.cw}@antgroup.com, cheng.yuan@fudan.edu.cn
Pseudocode No The paper describes the method using text and equations but does not include any explicit pseudocode or algorithm blocks.
Open Source Code Yes Our code is available at https://github.com/ant-research/Diverseand-Compact-Transformer.
Open Datasets Yes The proposed method is evaluated on four widely used person Re ID benchmarks, i.e., MSMT17 (Wei et al. 2018), Market-1501 (Zheng et al. 2015) and CUHK03 (Li et al. 2014).
Dataset Splits No The paper evaluates on well-known benchmarks (MSMT17, Market-1501, CUHK03) but does not explicitly provide the specific training, validation, and test splits (e.g., percentages or sample counts) for these datasets in the main text.
Hardware Specification Yes All the experiments are performed on 4 Nvidia Tesla V100 GPUs.
Software Dependencies No The paper mentions the use of a VIT-B/16 backbone and SGD optimizer but does not provide specific version numbers for software dependencies such as Python, PyTorch, or other libraries.
Experiment Setup Yes All the images are resized to 256 × 128 unless other specified. The training images are augmented with random horizontal flipping, padding, random cropping, random erasing (Zhong et al. 2020), and random grayscale (Gong et al. 2021). The initial weights of the models are pre-trained on Image Net. The batch size is set to 64 with 4 images per ID. SGD optimizer is employed with a momentum of 0.9 and a weight decay of 1e-4. The learning rate is initialized as 0.032 with cosine learning rate decay.