Adversarial Feature Disentanglement for Long-Term Person Re-identification

Authors: Wanlu Xu, Hong Liu, Wei Shi, Ziling Miao, Zhisheng Lu, Feihu Chen

IJCAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We evaluate our method on six person Re-ID datasets, including three large-scale long-term datasets: Celeb-re ID, Celeb-re ID-light, PRCC, one benchmark short-term dataset: Market-1501, and two collected long-term datasets: Market Clothes, PKU-Market-Reid which are shown in Fig. 3.
Researcher Affiliation Academia 1Key Laboratory of Machine Perception, Shenzhen Graduate School, Peking University, China 2Department of Precision Instrument, Tsinghua University, China
Pseudocode No The paper describes the proposed method using textual explanations and mathematical equations, but it does not include any pseudocode or clearly labeled algorithm blocks.
Open Source Code No The paper does not contain any explicit statements about making its source code publicly available or provide any links to a code repository.
Open Datasets Yes Moreover, we collect a challenging Market-Clothes dataset and a real-world PKU-Market-Reid dataset for evaluation. The results on one large-scale short-term dataset (Market-1501) and five long-term datasets (three public and two we proposed) confirm the superiority of our method against other state-of-the-art methods.
Dataset Splits Yes Celeb-re ID [Huang et al., 2020]... It has 20,208 images from 632 identities for training and 13,978 images from 420 identities for testing. PRCC [Yang et al., 2020]... The dataset has 150 and 71 identities for training and testing... Market-1501 [Zheng et al., 2015]... The dataset is split into two non-overlapping fixed parts: 12,936 images from 751 identities for training and 19,732 images from 750 identities for testing.
Hardware Specification Yes We implement the AFD-Net using Py Torch with only one NVIDIA GTX 1080Ti GPU.
Software Dependencies No The paper states "We implement the AFD-Net using Py Torch" but does not specify the version number of PyTorch or any other software dependencies.
Experiment Setup Yes All images are resized to 256 128 for input. The SGD optimizer is used to train Ei, Ec with lr = 0.002, momentum = 0.9. The Adam optimizer is applied to optimize G, D with lr = 0.0001, (β1, β2) = (0, 0.999). Following [Huang et al., 2018], we use a large weight λimg rec = 5 for the image reconstruction loss. The feature reconstruction loss is inaccurate in the initial stage, so we gradually increase λfea rec from 0 to 1 during the training process. Similarly, the identification loss Lrec id may make the training unstable due to the low quality of adversarial generated images at the beginning, so we set a small weight λrec id = 0.5. The remaining weights λraw id and λadv are set to 1.