Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Diverse Embedding Expansion Network and Low-Light Cross-Modality Benchmark for Visible-Infrared Person Re-identification

About

For the visible-infrared person re-identification (VIReID) task, one of the major challenges is the modality gaps between visible (VIS) and infrared (IR) images. However, the training samples are usually limited, while the modality gaps are too large, which leads that the existing methods cannot effectively mine diverse cross-modality clues. To handle this limitation, we propose a novel augmentation network in the embedding space, called diverse embedding expansion network (DEEN). The proposed DEEN can effectively generate diverse embeddings to learn the informative feature representations and reduce the modality discrepancy between the VIS and IR images. Moreover, the VIReID model may be seriously affected by drastic illumination changes, while all the existing VIReID datasets are captured under sufficient illumination without significant light changes. Thus, we provide a low-light cross-modality (LLCM) dataset, which contains 46,767 bounding boxes of 1,064 identities captured by 9 RGB/IR cameras. Extensive experiments on the SYSU-MM01, RegDB and LLCM datasets show the superiority of the proposed DEEN over several other state-of-the-art methods. The code and dataset are released at: https://github.com/ZYK100/LLCM

Yukang Zhang, Hanzi Wang• 2023

Related benchmarks

TaskDatasetResultRank
Cross-modality Person Re-identificationSYSU-MM01 (All Search)
Recall@174.7
142
Visible-Thermal Person Re-identificationRegDB Visible to Thermal
Rank-191.1
140
Cross-modality Person Re-identificationSYSU-MM01 (Indoor Search)
Rank-180.3
114
Visible-Infrared Person Re-IdentificationRegDB Thermal2Visible v1
Rank-1 Acc91.1
87
Visible-Thermal Person Re-identificationRegDB Thermal to Visible
Rank-189.5
79
Visible-Infrared Person Re-IdentificationSYSU-MM01 All Search v1
Rank-174.7
70
Visible-Infrared Person Re-IdentificationSYSU-MM01 (Indoor Search)
R180.3
42
Vehicle Re-identificationMSVR310
mAP28.68
29
Visible-Infrared Person Re-IdentificationSYSU-MM01 Indoor Search v1
Rank-180.3
27
Ship Re-identificationHOSS-ReID Optical to SAR
mAP31.3
19
Showing 10 of 23 rows

Other info

Follow for update