WHU Team ISVL receives the CVPR 2025 Champion Certificate.
Professor Zhi Gao's research team, ISVL, from the School of Remote Sensing and Information Engineering at Av性爱 (WHU) recently won first place in the 2025 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Visual Anomaly and Novelty Detection (VAND) Challenge. The team topped the "Robust Anomaly Detection in Real-World Applications" track, marking a major milestone in intelligent visual inspection.
Competition details and challenge significance
The VAND Challenge, first launched in 2023 as part of a CVPR workshop, has quickly become one of the most prestigious international competitions in the field of visual anomaly detection. This year's third edition specifically addressed the significant gap between academic research and real-world deployment.
In Challenge Track 1, titled "Adapt & Detect: Robust Anomaly Detection in Real-World Applications", Professor Gao's team claimed first place by addressing some core challenges in industrial visual inspection. A major technical challenge lay in accurately detecting microscopic defects that often appear on transparent or reflective surfaces under complex and variable lighting conditions. All participating methods operated within a fully unsupervised learning framework – trained solely on defect-free samples without prior knowledge of actual anomalies.
The Champion Certificate of the CVPR 2025 VAND Challenge (Team ISVL).
Technical achievements
The anomaly detection solution proposed by ISVL is of great significance to the field of industrial quality inspection. Capable of accurately identifying microscopic flaws on complex surfaces, it demonstrates outstanding robustness and provides a powerful intelligent tool for improving product quality. This achievement underscores the practical value and vast potential of artificial intelligence in industrial anomaly detection and highlights how AI-driven visual inspection can greatly optimize the quality assurance process.
Team ISVL presents a report at CVPR 2025.
Team profile and collaboration
ISVL achieved this top global ranking through dedicated research and weekly collaborative seminars under the close guidance of Professor Gao, who provided invaluable strategic direction and technical insights. Team members include PhD student Xingao Wang, master students Zhaohong Liao and Mengjie Xie, and undergraduates Shuying Xia and Handa Wang. The achievement demonstrates the real-world value and broad prospects of integrating photogrammetry and AI technologies in industrial anomaly detection.
Team ISVL's technical discussion session.
Expertise and vision
Professor Gao's team has been deeply engaged in the fields of intelligent unmanned systems and remote sensing anomaly detection for a long time. Their research focuses on the deep integration of advanced remote sensing science, photogrammetry, and AI, aiming to address technical challenges in real-world emergency response scenarios such as surface anomaly monitoring and industrial anomaly detection.
