• Colorization & Enhancement

• Based on the characteristics of images and system, an algorithm framework for video denoising is developed.

• Learned ISP processing, camera system, HDR fusion and noise calibration.

• Utilized PnP framework, which combined deep learning denoising method with traditional optimization method.

• Applied the self-supervised denoising method in the PnP framework to achieve a self-supervised video snapshot compressive imaging.

• Proposed a small object detection algorithm based on feature enhancement in airport video, which supplemented the features of small objects in a single frame image to provide a better image feature for subsequent temporal fusion.

• Developed two different Person Re-id networks according to spatial and temporal.