Papers › Image-to-image domain adaptation for vehicle re-identification
Image-to-image domain adaptation for vehicle re-identification
Fukai Zhang, Lulu Zhang, Haiyan Zhang, Yongqiang Ma
Cross-domain vehicle re-identification (ReID) is an interesting but challenging task in computer vision. A ReID model well-trained on one dataset often experiences a severe performance drop when applied to another dataset due to the domain discrepancy between the different datasets. This is especially true for low-resolution images. In this paper, we present a vehicle image domain adaptation framework (VDAF) which contains a single-image super resolution network (SISR) and a vehicle transfer generative adversarial network (VTGAN). SISR is an enhancement task for mapping low-resolution (LR) images to high-resolution (HR) images. Based on the reconstructed HR images, VTGAN can translate vehicle images from a source domain to a target domain with consistent styles and identities. VTGAN is an unsupervised approach designed for source-target translation for vehicle ReID and is composed of two adversarial networks and one Siamese network. Based on the translated images, we can infer an enhanced vehicle representation free of influences from style variations, allowing distance metrics for vehicle ReID to be learned. Through extensive experiments on the VeRi, VehicleID, and VRIC datasets, we show that images translated by VTGAN are effective for domain adaptation and are superior at promoting the accuracy of vehicle ReID.
Code
No code repository is listed for this paper in the archive or in Syntology's graph.
Code Syntology ran Syntology
Not run by Syntology. Nothing on this page verifies that the listed code works.
Tasks
1 archive task tag without a task page not shown.
Results from the paper archive 2025-07-28
Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.
Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections