FCA-ResNet: An Improved Model with Enhanced Multi-Scale Feature Fusion and Coordinate Attention for Wheat Leaf Disease Classification

Rapid and accurate identification of leaf disease is essential in intelligent agriculture. Current methods often struggle with balancing precision and speed. This research introduces the fusion coordinate attention and residual network (FCA-ResNet) model to improve classification accuracy while maintaining a lightweight structure for both healthy wheat leaves and five common wheat leaf diseases. FCA-ResNet incorporates a coordinate attention (CA) mechanism along with a multi-branch Inception module. The model consists of an Inception-based multi-branch structure and CA mechanism fusion module, which optimizes feature focus and weight allocation. Additionally, a multi-scale fusion module utilizes both channel and spatial attention mechanisms to effectively integrate shallow and deep features, improving the detection accuracy of small lesions. The multi-branch structure is designed to replace traditional multi-layer convolution, resulting in a lightweight model. The model achieves an average accuracy of 91.6% on custom datasets, demonstrating its effectiveness in plant disease detection for agriculture.