Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].

Base-Detail Feature Learning Framework for Visible-Infrared Person Re-Identification

Authors: Zhihao Gong, Lian Wu, Yong Xu

IJCAI 2025 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Comprehensive experiments conducted on the SYSUMM01, Reg DB, and LLCM datasets validate the effectiveness of BDLF.
Researcher Affiliation Academia Zhihao Gong1 , Lian Wu2 , Yong Xu1, 1Harbin Institute of Technology (Shenzhen) 2Gui Zhou Education University EMAIL, wulian EMAIL, EMAIL
Pseudocode No The paper describes the methodology using mathematical equations and textual explanations, but it does not contain any explicitly labeled pseudocode or algorithm blocks.
Open Source Code No The paper does not contain any explicit statement about releasing source code or provide a link to a code repository.
Open Datasets Yes Comprehensive experiments conducted on the SYSUMM01, Reg DB, and LLCM datasets validate the effectiveness of BDLF. SYSU-MM01 dataset [Wu et al., 2017] comprises 287,628 VIS and 15,792 IR images from 491 identities captured by 4 RGB and 2 IR cameras. It features both All-Search and Indoor-Search modes for evaluation. Reg DB [Nguyen et al., 2017] contains 412 identities, each represented by 10 VIS and 10 IR images captured from a pair of cameras. We adhere to the evaluation protocol outlined in [Ye et al., 2022b] to randomly split the identities into training and testing sets of equal size. LLCM [Zhang and Wang, 2023]is a challenging large-scale low-light dataset for VI-Re ID task, which contains 713 identities with 25,626 VIS and 21,141 IR images, all captured by 9 cameras in both RGB and IR modalities
Dataset Splits Yes We adhere to the evaluation protocol outlined in [Ye et al., 2022b] to randomly split the identities into training and testing sets of equal size. For each mini-batch, we randomly sampled 8 identities, each consisting of 4 VIS and 4 IR images for training.
Hardware Specification Yes The entire framework is implemented using Py Torch and runs on a single NVIDIA RTX3090 GPU with 24GB VRAM.
Software Dependencies No We employed a pre-trained Res Net-50[He et al., 2016b] as the backbone network and incorporated INN blocks with affine coupling layers[Dinh et al., 2017][Zhou et al., 2022] to construct the DFE module, setting the number of INN blocks to 6. The SGD optimizer was used
Experiment Setup Yes The SGD optimizer was used, with the initial learning rate set to 1 10 2, which was warmed up to 1 10 1 during the first 10 epochs, then we decayed the learning rate to 1 10 2 and 1 10 3 at epochs 20 and 95 for SYSU-MM01, and at epochs 70 and 130 for Reg DB and LLCM, respectively. The learning rate was further decayed to 1 10 4 at 180 epoch, with a total of 220 epochs. For each mini-batch, we randomly sampled 8 identities, each consisting of 4 VIS and 4 IR images for training. Additionally, the exponential moving average (EMA) model [Ge et al., 2020] also employed in our method.