Frame labeling-based weak supervision video target segmentation method and device

• Main Authors: WANG JUN, TAN CHAOLEI, ZHENG WEISHI, HU JIANFANG, LIN ZIHANG
• Format: Patent
• Language: Chinese; English
• Published: 07.03.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; PHYSICS

The invention discloses a frame annotation-based weak supervision video target segmentation method and device. The method comprises the following steps of training a PReMVOS model-based pseudo annotation generation model on an image segmentation data set; generating corresponding pseudo mask labels for the video data and the frame labels frame by frame by using a pseudo label generation model; and training a video target segmentation model by using a cooperative teaching algorithm by using the generated pseudo mask label. According to the method, the video target segmentation model is trained by using low-cost frame labeling, so that the cost of labeling new data when the video target segmentation model is migrated to an actual application scene is reduced, and the landing difficulty of the video target segmentation model is reduced; by using a training algorithm of cooperative teaching, a large number of existing video target tracking data sets can be fully utilized to train a video target segmentation model