Computer vision enables computers to interpret and analyze visual data such as images and video. Deep learning applications commonly use convolutional neural networks to process that data. section 8.0
Most practical computer vision systems address one of three foundational tasks:
- Image classification assigns one or more labels to an image, such as identifying a cat or dog. section 11.1
- Image segmentation partitions an image into labeled regions with pixel-level precision. section 11.1
- Object detection draws bounding boxes around objects and associates each box with a class. section 11.1
From photos to autonomous systems
Google Photos can use multi-label classification to answer image searches; video-call backgrounds use segmentation; and autonomous vehicles can use detection to monitor cars, pedestrians, and signs. Vision models also support OCR, robotics, medical diagnosis, retail checkout, and farming. section 8.0 section 11.1
Figure 11.1 The three main computer vision tasks: classification, segmentation, and detection
Compare classification, segmentation, and detection as the three central computer vision tasks. section 11.1
