Identification of Apple Leaf Diseases by Improved Deep Convolutional Neural Networks With an Attention Mechanism

Accurate identification of apple leaf diseases is of great significance for controlling the spread of diseases and ensuring the healthy and stable development of the apple industry. In order to improve the detection accuracy and efficiency, a deep learning model, which called CA-ENet, is proposed to identify different apple diseases. First, coordinate attention block is intergrated into the EfficientNet-B4 network, embedded the feature's spatial location information by channel attention to ensure that the model can learn both channel and spatial location information of the important features. The depthwise separable convolution is applied to the convolution module to reduce the number of parameters, and the h-swish activation function is introduced to achieve fast and easy to quantify the process. 5,170 images are collected in the field environment at the apple planting base of Northwest A&F University, while 3,000 images are acquired from the PlantVillage public data set. Also image augmentation techniques are used to generate an Apple Leaf Disease Identification Data set (ALDID), which containing 81,700 images. The experimental results show that the accuracy of the CA-ENet is 98.92% on the ALDID, and the average F1-score reaches 0.988, which is better than common models such as ResNet-152, DenseNet-264 and ResNeXt-101. The generated test data set is used to test the model's anti-interference ability. The results show that the proposed method can achieve competitive performance on apple disease identification task.