Yichen Lu, Siwei Nie, Minlong Lu, Xudong Yang, Xiaobo Zhang, Peng Zhang
The paper introduces PixTrace and CopyNCE to improve image copy detection by enhancing pixel-level traceability and patch-level similarity learning, achieving state-of-the-art performance on the DISC21 dataset.
This research focuses on improving how we detect copied and altered images. The authors developed two new methods: PixTrace, which keeps track of pixel positions even after an image is edited, and CopyNCE, which helps the system learn better by using information about how image patches overlap. These innovations make the detection system more accurate and easier to understand, outperforming previous methods in tests.
