SJDL-Vehicle: Semi-supervised Joint Defogging Learning for Foggy Vehicle Re-identification

Vehicle re-identification (ReID) has attracted considerable attention in computer vision. Although several methods have been proposed to achieve state-of-the-art performance on this topic, re-identifying vehicle foggy scenes remains a great challenge due the degradation of visibility. To our knowledge, problem is still not well-addressed so far. In paper, address problem, we propose novel training framework called Semi-supervised Joint Defogging Learning (SJDL) framework. First, fog removal branch and are integrated perform simultaneous training. With collaborative scheme, defogged features generated by defogging from input images can be shared learn better representation for branch. However, since fog-free image real-world data intractable, architecture only trained synthetic data, which may cause domain gap between scenarios. solve design semi-supervised scheme that train two kinds alternatively each iteration. Due lack dataset specialized ReID weather, construct FVRID consists evaluate performance. Experimental results show method effective outperforms other existing weather. The code available https://github.com/Cihsaing/SJDL-Foggy-Vehicle-Re-Identification--AAAI2022.