Research on the Construction of Fine-Grained Recognition Model of Genuine Medicinal Materials Images Based on GhostNet

With the rapid development of artificial intelligence, image recognition technology has been widely concerned in the field of traditional Chinese medicine. In particular, the deep learning method represented by convolutional neural network performs well in image classification tasks, which provides a new idea for image recognition of Chinese medicinal materials. However, most of the existing deep learning models have some problems, such as strong sample dependence, high computational complexity and large number of parameters, which are difficult to meet the actual project application requirements. In this regard, this paper will deeply explore the feasibility of different types of technical solutions, and put forward a lightweight, efficient and accurate fine-grained recognition model of genuine Chinese medicinal materials images, so as to improve the quality control level of Chinese medicinal materials and ensure the safety of medication. Practice has proved that the whole model is built based on GhostNet algorithm, and Ghost feature generation and multi-scale feature fusion are realized through Ghost Bottleneck module to improve the running efficiency and recognition accuracy of the model, so as to further improve the detection performance of the model.