What is the principle of product semantic subdivision?

As a key task in the field of computer vision, semantic segmentation is the only way to realize complete scene understanding. In order to make the machine have vision, it must go through the process of image classification, object detection and image segmentation. Among them, image segmentation is the most difficult.

More and more applications benefit from image classification and segmentation technology, and full scene understanding is also very important in the field of computer vision. Some of these applications include self-driving vehicles, human-computer interaction, AR-VR and so on. With the popularity of deep learning in recent years, many semantic segmentation problems are solved by deep architecture, among which CNN (Convolutional Neural Network) is the most common, and its accuracy and efficiency greatly exceed other methods.

What is semantic segmentation?

Semantic segmentation is a natural step of reasoning from coarse to fine: the origin can be located in classification, including predicting the whole input. The next step is location/detection, which not only provides classes, but also provides additional information about the spatial location of these classes. Finally, semantic segmentation realizes fine-grained reasoning by densely predicting the reasoning label of each pixel, so each pixel is marked as its category around the object ore area.

More specifically, the goal of semantic image segmentation is to mark each pixel in the image with corresponding kinds of content. Because we want to predict every pixel in the image, this task is usually called dense prediction. It should be noted here that semantic segmentation is the task of assigning a category to each pixel in a given image, rather than assigning a single category to the whole image. Semantic segmentation is to classify each pixel in an image into one of these categories.

Application of Semantic Segmentation

Self-driving: It is necessary to equip the car with the necessary perception ability, so that the car can "observe" the road conditions and surrounding environment, so that the self-driving car can drive safely on the road.