Research

1. Depth Perception
This research is focused on: depth completion, which involves filling in missing data in partial depth maps; depth estimation, leveraging video sequences to ascertain depth values for each frame; and depth super-resolution, which upscales low-resolution depth maps to a higher resolution to match the clarity of high-resolution RGB images.

2. Creative Image Generation Based on Text-to-Image Models
This primarily involves creative image generation from text, creative image editing, and creative 3D object generation, all of which have strong applied research value and high industrial potential.

3. Real-world Moving-Video Dehazing and Enhancement
This advancement aims to bolster visibility for drivers in low-visibility situations and to elevate the quality of footage for video monitoring systems.