Improve of Mask R-CNN in Edge Segmentation ☆

Nowadays, grasping robot plays an important role in many automatic systems in industrial environment. An excellent grasping robot is which can detect, localize, and pick objects accurately. However, to perfectly achieve these tasks, it is still a challenge in computer vison field. Especially, segmentation task, which is understanded as both detection and localization, is the hardest problem. To deal with this problem, the state-of-the-art Mask R-CNN was introduced and obtained an exceptional result. But this superb model does not certainly perform well when working with harsh location of objects. The edge and border regions are usually misunderstood as the background, this leads to the failure in localizing object to submit a good grasping plan. Thus, in this paper, we introduce a novel method that combine original Mask R-CNN pipeline and 3D algorithms branch to preserve and classify the edge region. This results the improvement of the performance of Mask R-CNN in detailed segmentation. Concretely, the significant improvement practiced in harsh situation of object location obviously discussed in experimental result section, especially edge regions. Both IoU and mAP indicator are increased. Specifically, mAP, which directly reflects the semantic segmentation ability of a model, raised from 0.39 to 0.46. This approach opens a better way in determine the object location and grasping plan.