Abstract: Aiming at the problem of high complexity in mask representation in the field of instance segmentation, we proposed a new mask representation method for instance segmentation, which used three repsesentation units that did not rely on any prior information  to represent and predict mask, and restored the mask in the form of nonlinear decoding. This method could significantly reduce the representation complexity and inference computation of image instance masks.   Based on the representation method, we constructed an efficient single-shot instance segmentation model. The experimental results show that compared to other single-shot instance segmentation models, the model can achieve better performance while ensuring that the  time cost is basically the same. Additionally, we embed the representation method with minimal modifications into the classic model BlendMask to reconstruct attention maps. The improved model has a  faster inference speed compared  to the original model, and the average accuracy of the mask is improved by 1.5%, indicating that the  representation method has good universality.

Key words: deep learning, instance segmentation, compressed representation, representation unit