Instance Segmentation

Instance segmentation is a computer vision task that involves detecting and delineating each individual object instance within an image at the pixel level. This task extends traditional object detection by not only identifying the bounding boxes of objects but also providing precise pixel-wise masks for each instance. Mathematically, instance segmentation can be formulated as a combination of object detection and semantic segmentation, where the model outputs a set of masks corresponding to detected objects. Popular architectures for instance segmentation include Mask R-CNN, which employs a region-based convolutional neural network (R-CNN) to generate region proposals and subsequently predicts binary masks for each instance. The evaluation of instance segmentation models is typically performed using metrics such as Average Precision (AP) and Intersection over Union (IoU).