AUTHOR=Fu Lili , Li Shijun , Sun Yu , Mu Ye , Hu Tianli , Gong He TITLE=Lightweight-Convolutional Neural Network for Apple Leaf Disease Identification JOURNAL=Frontiers in Plant Science VOLUME=Volume 13 - 2022 YEAR=2022 URL=https://www.frontiersin.org/journals/plant-science/articles/10.3389/fpls.2022.831219 DOI=10.3389/fpls.2022.831219 ISSN=1664-462X ABSTRACT=As one of the best-selling fruits, it is extremely important to control fruit tree diseases on apples. In this research, we design a lightweight convolutional neural network based on Alexnet model for 5 diseases of apple leaves. Firstly, the use of dilated convolution in the model to extract coarse-grained features to reduce the number of parameters while maintaining a receptive field. then a multiscale module was added to replace the original 5×5 convolution kernel to extract leaf lesions at multiple scales. 3 × 3 convolutional short-circuiting of serial connections to give the model more nonlinearity. addition of attention mechanisms behind all modules of aggregated outputs for better fitting of channel features. Finally, the two fully connected layers are replaced with global pooling to reduce the number of model parameters without losing features. The final recognition accuracy of the model is 97.36%, which is 9.31 percentage points better than the original Alexnet. The performance of the model is verified and compared with other 4 networks, and the results show that the size of our model is 5.87MB, which is 2.64MB less than the params size of Mobilenet, which is famous for its light weight. This research provides the possibility of migrating disease identification models to mobile devices and provides a reference for further research on apple leaf disease identification methods.