Adaptive feature fusion-based approach to tomato leaf disease identification
Image classification techniques were widely used in agriculture,especially in disease detection and classification,and were found to be more efficient and accurate than traditional manual methods.Traditional feature fusion methods used fixed weighting operations to enhance local features and suppress the expression of interfering features,but the differences in disease class images affected the generalization ability of the model,resulting in lower classification efficiency and accuracy.In order to address this issue,a tomato leaf disease recognition method based on multi-layer adaptive feature fusion was proposed in this study.The data set was first enhanced by a data enhancement algorithm to alleviate the problems of insufficient data sample size and category imbalance.Then,feature enhancement was used to capture key features,followed by feature fusion with adaptive weights,resulting in accurate recognition of tomato leaf disease categories.The proposed method achieved a recognition accuracy of 99.67%for tomato leaf disease images,which was an improvement of 2.07%-15.33%compared to other deep network models such as InceptionV3 and ResNet50.Accurate image recognition of tomato leaf diseases was achieved by using this method,which provided ideas and methods for the recognition technology of tomato and other crop diseases.
data enhancementtomato leaf diseaseimage classificationfeature fusionfeature enhancement
NETL
NSTL
万方数据
