2021 IEEE International Conference on Real-time Computing and Robotics (RCAR) | 2021
Rapid identification method of fresh tea leaves based on lightweight model
Abstract
A Lightweight Convolutional neural network mode (MobileNetV2-Tea) was designed based on computer vision technology, given the difficulty in accurately subdividing fresh tea leaves by using traditional sorting methods. By improving the MobileNetV2 network, the obtained MobileNetV2-Tea model has an accuracy of 99% in the recognition of fresh tea images. The experimental results show that the MobileNetV2-Tea model has a higher average accuracy of 2 to 5 percentage points than other models. The model size is only 28.86M, and the average recognition time is 45ms. Therefore, the recognition model has the advantages of small size, high recognition accuracy, fast recognition time. It meets the requirements of the model for industrialized embedded devices. So the fresh tea leaves sorting technology based on the lightweight convolutional neural network model (MobileNetV2-Tea) solves the problem of relying on manual sorting and mixing of different types of fresh tea, and improves the quality and value of teas.