Ali won four World Championships in CVPR 2020, and new research on pixel video segmentation


Recently, CVPR 2020, the global computer vision summit, announced the results of major challenges. Ali won the world championship in four competitions. The new research in the competition will be applied to live broadcast, online conference and other scenes on a large scale.
CVPR is the largest conference in the field of AI, covering prospective academic research and industrial applications in the field of computer vision. It is understood that Ali provides a new spatial constraint method, which breaks the bottleneck of the lack of timing in traditional STM method, and enables the system to predict the position of the next frame of the target object based on the image of the previous frame of the video; in addition, Ali introduces the fine segmentation and fine adjustment module in semantic segmentation, which greatly improves the precision of segmentation. Finally, the contour boundary of the dynamic target is accurately identified and separated from the background to achieve pixel level target segmentation.