The picture of the same person holding the ID card means taking a photo of the upper body of the person holding the ID card.
The goal of semantic understanding of portrait pictures is to obtain the category label of each pixel in the portrait picture (such as face, hair, skirt, top, etc.). This task involves image classification, object detection and semantic segmentation The core technology of computer vision is a very challenging research topic and a major basic issue that has a wide impact on society and economy.
On the other hand, this solution takes into account the contextual relationship between different semantic tags and can also meet the real-time needs of the actual system. This project intends to make a breakthrough in the task of semantic understanding of portrait pictures, and in It has achieved internationally leading research results in terms of paper publication, patent application and talent training.