In order to accurately sort the impurities and defective materials in the raw Chinese medicinal materials,an improved YOLOv6 algorithm for Chinese medicinal materials sorting is proposed to solve the difficult detection of the materials with inconspicuous defective feature.First,based on the characteristics of medicinal materials,the redundant large target detection head of the detection network is cut to reduce false detections;Then,the SPD convolution is introduced instead of strided convolution for image downsampling to enhance the ability to extract fine grained features.Finally,the coordinate attention mechanism is introduced in the network to improve the attention of important features.The experiments are conducted based on the astragalus dataset.The results show that the mAP of the improved YOLOv6 algorithm is 86.2%,which is 2.9%higher than the original algorithm and has the stronger detection ability for small defective medicinal materials.
With the advancement of urbanization,the traffic situation in cities has become increasingly congested,and how to dispatch and manage the traffic flow has become an urgent problem to be solved.By designing an efficient and intelligent traffic control system,the problem of traffic congestion can be greatly reduced.However,at present,the traffic lights on most road sections are still timed based on specific schedules.This control method works well when the traffic flow is low,but it cannot meet the demand due to large traffic flow or changes in traffic conditions.Therefore,this article designs an intelligent traffic control system that collects traffic flow information,calculates the corresponding relationship between intersection traffic flow and passing time using the BP neural network control algorithm,and improves the algorithm model through independent learning and training of the system,thereby achieving the goal of intelligent adjustment of traffic flow.This system designs a control system model based on BP neural network for intelligent transportation,which predicts the passage time of various types of transportation vehicles.Through this design,the traffic light control system can be more intelligent,with certain economic and social benefits.
In this paper,Citespace uses the diffusion model as the title and the documents dated from 2012 to 2022 to conduct data analysis.It objectively analyzes the number of documents,countries,institutions,keywords,references and other aspects of the diffusion model documents,and obtains that the number of documents issued by the diffusion model has been on the rise,indicating that the research on the diffusion model has been a hot topic for scholars in recent years.Through the analysis of the distribution of countries and institutions,as well as the number of papers issued,it is found that China and the United States have more research on the diffusion model,and the lines between countries are closely connected,which indicates that the academic exchanges between countries on the diffusion model are more extensive,and the cooperation between institutions is also closer.Among them,the Chinese Academy of Sciences has the largest number of papers issued,while some universities and institutions in the United States have published more.Through the analysis of keyword co-occurrence,clustering,highlighted words and references,it can be seen that the diffusion model in 2012-2022 is a hot topic in the field of medical research.In combination with other clustering tagwords,it can be seen that the diffusion model is also widely studied in the ecological field,and in infectious diseases,machine learning is also widely favored by scholars.
Aiming at the problems of vehicle detection by YOLOv5 model in complex traffic scenarios,such as slow detection speed,serious error detection,and difficulty in small object recognition,a new model YOLOv5-Ours with fast model convergence speed and high target frame accuracy is proposed.Firstly,the Bidirectional Feature Pyramid network(BiFPN)structure is used to improve the recognition accuracy by using the fine-grained features of multi-scale small objects,so as to mine the fine-grained features of different images.Secondly,the composite structure of the Spatial Pyramid Pool Fast(SPPF)structure and the ELU activation function is used to achieve fast and accurate target detection.The experimental results show that,compared with other algorithms,the algorithm has faster speed,higher precision and good robustness.
Session recommendations were designed to predict the next item they will interact with based on a sequence of historical user behavior.How to deeply analyze the complex dependencies within the conversation sequence and accurately extract the potential preferences of users is a huge challenge in the design of the current conversation recommendation model.In view of this,a graph neural network session recommendation model(AMSR-GNN)combined with attention mechanism was proposed.Firstly,all session data was constructed as a graph,and the local embedding representation of nodes on the graph was obtained through the graph neural network.Secondly,the attention mechanism containing the noise filter was used to display and filter out the unimportant node representations to obtain the denoising-enhanced global embedding representation,and finally,the local embedding representation and global embedding representation of the item are considered by the prediction layer to generate personalized recommendations for users.
