HLNet Model and Application in Crop Leaf Diseases Identification

Crop disease has been a severe issue for agriculture, causing economic loss growers. Thus, identification urgently needs to be addressed, especially precision agriculture. As of today, deep learning widely used crop combined with optical imaging sensors. In this study, lightweight convolutional neural network model is designed and validated on two publicly available datasets one self-built dataset 28 types leaf images 6 crops as the research object. This an improvement existing network, reducing floating-point operations by 65%. addition, dilated depth-wise convolutions were increase receptive field improve recognition accuracy without affecting computational speed. Meanwhile, attention mechanisms are optimized reduce module computation, improving capability select correct regions interest. After training, achieved average 99.86%, image calculation speed was 0.173 s. Comparing 11 backbone models 5 latest studies, proposed highest accuracy. Therefore, advantage balancing between Furthermore, provides theoretical basis technical support practical application mobile terminal applications in