Semantic Segmentation

A computer vision technique that classifies every pixel in an image into a predefined category, producing a detailed map of what objects are present and where they are located. Unlike object detection (which draws bounding boxes around objects), semantic segmentation provides precise boundaries for each region. In accessibility applications, semantic segmentation enables systems to identify individual items within complex images — for example, distinguishing a top, pants, bag, and shoes in a clothing photograph — allowing blind or low vision users to query specific items rather than receiving a single generic image description. It also supports smart magnification by cropping and enlarging only the item of interest. Semantic segmentation is typically implemented using convolutional neural networks trained on large annotated datasets.