Research on an Automatic Tomato Pest and Disease Recognition Algorithm Based on ResNet50

Tomatoes are highly nutritious, and as the scale of their cultivation continues to expand, accurately and efficiently identifying pests and diseases is crucial for their management. Traditional methods for identifying tomato pests and diseases primarily rely on manual techniques, which are not only inefficient and inaccurate but also costly. The automatic identification of tomato pests and diseases based on deep learning can achieve accurate and efficient detection, offering significant practical value. Building on existing networks, we proposed an automatic tomato pest and disease identification method based on ResNet50. First, the pre-trained ResNet50 is used to extract global features from images of tomato pests and diseases. Then, an efficient attention mechanism module is added in series to the network, allowing the model to effectively integrate global and local feature information, thereby improving identification accuracy. Experimental comparisons with existing classical network models show that our proposed model achieves a recognition accuracy of 98.85%, demonstrating significant advantages.