Corn Disease Recognition Based on Attention Mechanism Network

To extract more accurate and abundant features of corn disease solve the problems rough classification low recognition accuracy, attention mechanism is introduced into field recognition. The model (AT-AlexNet) proposed based on an mechanism. network was AlexNet, new down-sampling module constructed to enhance foreground response disease; Mish activation function improve nonlinear expression network; full connection layer designed reduce parameters. In experiment enhanced datasets, average accuracy attention-based AT-AlexNet 99.35%. using 0.65% higher than that ReLu function. experiments show compared with other identification methods, method has better performance for diseases.