A CNNA-Based Lightweight Multi-Scale Tomato Pest and Disease Classification Method

Tomato is generally cultivated by transplanting seedlings in ridges and furrows. During growth, there are various types of tomato pests diseases, making it challenging to identify them simultaneously. To address this issue, conventional convolutional neural networks have been investigated, but they a large number parameters time-consuming. In paper, we proposed lightweight multi-scale pest disease classification network, called CNNA. Firstly, constructed dataset diseases consisting 27,193 images with 18 categories. Then, compressed optimized the ConvNeXt-Tiny network structure maintain accuracy while significantly reducing parameters. addition, feature fusion module improve extraction ability model for different spot sizes pests, global channel attention mechanism enhance sensitivity features. Finally, was trained deployed Jetson TX2 NX inference video stream data. The experimental results showed that CNNA outperformed pre-trained models such as MobileNetV3, MobileVit, ShuffleNetV2 terms all parameters, recognition 98.96%. Meanwhile, error rate, time single image, FLOPs, size were only 1%, 47.35 ms, 0.37 M, 237.61 1.47 MB, respectively.