Distance to Center of Mass Encoding for Instance Segmentation
This work addresses instance segmentation for computer vision applications, presenting a new annotation encoding method that could improve object delineation and occlusion handling.
The authors tackled instance segmentation by proposing a novel encoding technique called Distance to Center of Mass Encoding (DCME), which represents instances with a center of mass and vector fields, enabling deep semantic segmentation models to learn and generalize this representation.
The instance segmentation can be considered an extension of the object detection problem where bounding boxes are replaced by object contours. Strictly speaking the problem requires to identify each pixel instance and class independently of the artifice used for this mean. The advantage of instance segmentation over the usual object detection lies in the precise delineation of objects improving object localization. Additionally, object contours allow the evaluation of partial occlusion with basic image processing algorithms. This work approaches the instance segmentation problem as an annotation problem and presents a novel technique to encode and decode ground truth annotations. We propose a mathematical representation of instances that any deep semantic segmentation model can learn and generalize. Each individual instance is represented by a center of mass and a field of vectors pointing to it. This encoding technique has been denominated Distance to Center of Mass Encoding (DCME).