SE-PSNet: Silhouette-based Enhancement Feature for Panoptic Segmentation Network

In this work, we propose a panoptic segmentation model that integrates bottom-up and top-down methods. Our framework is designed to guarantee both the performance inference speed. We also focus on improving quality of semantic instance masks. The proposed auxiliary task with silhouette-based enhanced features can help improve prediction mask contours. Additionally, introduce new score intending solve occlusion problem. has less chance ignoring small objects, which often have lower confidence scores than larger objects behind them. results show better distinguish priority when occurs. demonstrate our work two datasets: COCO dataset CityScapes dataset. Via approach, obtained competitive fast time.