Algae Image Classification Algorithm Based on the Improved MobileNetV2

Abstract

To address the problems of large number of parameters, poor real-time performance and low classification accuracy of existing algae image classification models, this paper proposes a lightweight model based on MobileNetV2. By using the GELU activation function instead of the RELU activation function, the generalization ability of the model and classification accuracy are improved; In order to establish the dependency between channel information and location information, a lightweight coordinate attention mechanism is embedded in the model. The experimental results show that the model can efficiently identify algae categories, and the overall recognition accuracy of the model reaches 97.0% on the algae image dataset after convergence. Moreover, the number of model parameters is only 10.68M, which has certain practical application value.