Attention aware cross faster RCNN model and simulation

Abstract

The rapid development of artificial intelligence has made various automated machines generally appear. In the catering industry, automatic billing has great development prospects. At present, the commonly used recognition methods include food recognition and dish recognition. There are many types of Chinese food, which are not conducive to promotion. Therefore, this article focuses on dish recognition. Faster R-CNN is a model with better results. There are also many improved networks to apply for the identification of various situations. This article improves Faster R-CNN, combining Faster R-CNN with cross-connected layers, and proposes the Cross Faster R-CNN model, which combines low-level features and high-level features, and introduces an attention mechanism to make the model highlight the characteristics of the dishes. The experimental results show that the Cross Faster R-CNN model, which introduces the cross-connected layer and attention mechanism, has no major changes in detection speed, and the accuracy is significantly improved.