Research On Aesthetics-based Image Cropping Technology

As one of the main sources of agricultural disasters,crop diseases and insect pests cause high food crop losses and economic losses every year in the world,and millions of people are also facing the crisis of undernourishment.In view of such a severe situation,it is of great research significance and application value to carry out intelligent classification and identification of crop diseases and insect pests for timely prevention and control.This paper studies the classification of crop diseases and insect pests based on images.The main research results include:1.This paper proposes a multi-scale fine-grained recognition algorithm based on the saliency of deep learning and attention mechanism.On the basis of traditional image features,the algorithm adds the salient features extracted by deep neural network to increase the level of feature representation;at the same time,for the characteristics of different lesion sizes,multi-scale convolution kernel is introduced to improve The stability of the algorithm to multi-scale changes;Finally,by introducing the attention mechanism,the algorithm’s attention to the lesions is further strengthened.2.Based on the aforementioned algorithm,this paper designs and implements a crop pest identification system.The system not only supports users ’online classification requirements for single images,but also supports users’ requirements for classification of a large number of images at the same time.The user selects the demand method on the front-end user interface and uploads the picture,and the back-end accepts the picture uploaded by the user,and obtains the final classification result through the trained model,and returns the top5 classification result to the interface,and gives statistical data,so that users have a deeper understanding of the situation of pests and diseases.Experimental results show that the algorithm in this paper is more reasonable,and the system design structure is clear and easy to use.