Instance Segmentation of Irregular Deformable Objects for Power Operation Monitoring Based on Multi-Instance Relation Weighting Module

Electric power operation is necessary for the development of grid companies, where safety monitoring electric difficult. Irregular deformable objects commonly used in electrical construction, such as belts and seines, have a dynamic geometric appearance which leads to poor performance traditional detection methods. This paper proposes an end-to-end instance segmentation method using multi-instance relation weighting module irregular objects. To solve problem introducing redundant background information when horizontal rectangular box detector, Mask Scoring R-CNN perform pixel-level so that bounding can accurately surround Considering workplaces often appear with construction personnel apparent correlation, proposed fuse features between are learned improve effect The mAP on self-built dataset reached up 44.8%. With same 100,000 training rounds, improved by 1.2% 0.2%, respectively, compared MS R-CNN. Finally, order further verify generalization practicability method, intelligent system scenes designed realize actual deployment application method. Various tests show segment well.