Instance segmentation is a computer vision task that combines object detection and semantic segmentation by identifying and segmenting each object instance in an image at the pixel level. This means not only classifying each pixel into a category, but also differentiating between separate objects of the same class. For example, in an image with three people, semantic segmentation would label all person pixels as “person,” whereas instance segmentation would assign person-1, person-2, person-3 different masks. A prominent model for instance segmentation is Mask R-CNN, which extends Faster R-CNN (for detection) by adding a branch that outputs a segmentation mask for each detected object. Challenges include ensuring masks precisely align with object boundaries and handling occlusions (where one object partly covers another). Instance segmentation output is often visualized by coloring each object mask differently. This task is important in applications requiring detailed scene understanding, like robotics (grasping specific objects) or medical imaging (segmenting individual cells or lesions).