Professor Gao Zhi's research team from the School of Remote Sensing and Information Engineering at Wuhan University in Hubei province recently won first place at the visual anomaly and novelty detection challenge of IEEE/CVF Conference on Computer Vision and Pattern Recognition 2025. The team topped the Robust Anomaly Detection for Real-World Applications category.

The challenge was first launched in 2023 as part of a CVPR workshop. It has become one of the most prestigious international visual anomaly detection competitions. This year's edition addressed the gap between academic research and real-world deployment.

Gao's team claimed first place by addressing a technical challenge in accurately detecting microscopic defects that often appear on transparent or reflective surfaces under complex and variable lighting conditions.

All participating methods were required to operate within a fully unsupervised learning framework using defect-free samples with no prior knowledge of anomalies.

The breakthrough anomaly detection solution proposed by Gao's team is of great significance in the field of industrial quality inspection. This technology can accurately identify microscopic flaws on complex surfaces, demonstrating outstanding robustness and providing a powerful tool for improving product quality, according to the university.

This achievement demonstrates AI's practical value and vast potential in industrial anomaly detection.