| Summary: | In this study, convolutional neural network (CNN) was used to classify and recognize the strawberry diseases and pests which cause tremendous economic losses for farmers. Based on AlexNet, the goal of this paper is to improve the recognition algorithm of strawberry diseases and pests by using the following methods: First, transfer learning is used to accelerate the time of model training and improve the recognition accuracy. Second, the model is fine-tuned to reduce the amount of training parameters and training time. The operator of combing inner product and ℓ∞ -norm was proposed and applied in the last fully connected layers to further improve the accuracy of recognition for strawberry diseases and pests. Compared with other start-of-the-art models, the experimental results show that our model achieves lower computational complexity and higher top-1 accuracy of the recognition for the strawberry diseases and pests on the test set.
|
|---|