Rapid identification method based on fine-grained image classification

Abstract

The invention discloses a rapid identification method based on fine-grained image classification, and belongs to the technical field of deep learning and image classification, and the method comprises the steps: a target picture is crawled from a network, a data set is made, and the network training and testing is carried out, only a half area or even a quarter area in one picture is actually used for fine-grained image classification, other parts are background areas and interference noise, contribution degrees of different areas in the picture to classification results are different, a Grad-CAM attention mechanism is used for obtaining a high-contribution-degree area and a low-contribution-degree area, framing and cutting are carried out, feature fusion is carried out on the high-contribution-degree area and the low-contribution-degree area, two calculation results are sent to a contribution degree module at the same time, and classification results are calculated through different contribution degree influence values. The application blank of fine-grained image classification in the field of rapid identification is filled.

Description

Technical field [0001] The present invention relates to depth learning and image classification technologies, in particular, a rapid identification method based on fine granular image classification. Background technique [0002] In recent years, the rapid development of artificial intelligence technology, the neural network and computer visual research have achieved significant breakthroughs, and various new networks and their improved are proposed, and the accuracy of fine-grained image classification is therefore continuously improved. As the accuracy improves, fine-grained image classification technology is widely put into practical applications, which greatly facilitates people's lives and work. The fine-grained image classification application is very wide. In terms of animal protection, fine-grained image classification can automatically identify which type is automatically identified, and does not require human resources and reduce human resources and reduce human eye rec...

AI Technical Summary

Problems solved by technology

[0003] However, the application of fine-grained image classification technology in the field of medical devices is very scarce. The main reasons are as follows: 1. Fine-grained image classification technology is in a stage of rapid development, with few research results and low accuracy, and the technical status is unstable; 2. Medical care is a major event related to life, so the requirements for accuracy are quite high. Before that, the accuracy of fine-grained image classification could not meet the standard

Images