According to statistics, the accident rate caused by fatigue driving and mental inconcentration in traffic accidents accounts for 93%. To solve this problem, 68 key points of face are introduced in this paper. Based on deep learning and machine vision algorithm concepts, combined with YOLOv8 model and related fatigue driving judgment mechanism, PERCLOS algorithm, MAR algorithm, EAR algorithm and HPE algorithm are used to improve the accuracy and reliability of the system. A set of algorithms for recognizing tired driving behavior is successfully constructed.

In order to alleviate the problem of seat tension in university canteens, it is urgent to develop a kind of software to solve this problem and improve the dining experience for students. This article discusses the use of the current mainstream microservice framework Spring Boot, combined with technologies such as MyBatis Plus and Vue.js, to design and implement a campus ordering system backend, which facilitates platform administrators and merchant administrators to manage dishes, packages, orders, etc., providing a solution for the problem of tight dining for college students.

Helmet detection is a computer vision task with important application value, involving safety management in many fields such as construction sites, mines, and electric power. However, helmet detection also faces many challenges, such as large changes in target size and aspect ratio, rapid changes in target velocity, target occlusion, and background interference. In order to solve these problems, this paper proposes a safety helmet detection method based on YOLOv8, which uses the characteristics of high speed and high precision of YOLOv8 combined with the characteristics of safety helmets to achieve effective detection and identification of safety helmets.

With the rapid development of information technology and the growing demand for financial management and financial analyses, bookkeeping is no longer limited to ordinary handwriting or manual recording in software. In order to meet the needs and development of today's era, an intelligent financial system based on OCR and large language model is designed. The system is suit‐ able for mobile phones, with financial management functions as the starting point and simplified management operations as the pur‐ pose, and realizes a system that is mainly based on OCR recognition scanning and recording bookkeeping functions, supplemented by financial analysis functions of big language model, big data consumption classification trends, and multi-version switching of ver‐ sions suitable for people of different age groups. The system not only greatly improves the efficiency of people's bookkeeping, but also promotes the intelligent financial transformation of society.

An improved algorithm is proposed to address the problem that yolov5 detects UAV aerial images with poor monitoring performance for low light background targets. Firstly, data normalization is performed on the target dataset visdrone2019 to improve the detection effect. Then the dynamic convolution kernel with Mish activation function and the C3_DSConv module using distrib‐ uted offset convolution to replace the C3 block are introduced, and the above two convolution structures are fused into the yolov5 network; the BiFormer attention mechanism is embedded to improve the accuracy of small target detection. In summary, the MODByolov5 model is finally obtained, and the experimental results prove that the model's mAP and recall are both improved, and the ac‐ curacy of detecting vehicles in shadows and dark environments is significantly increased, and the FPS is high, which ensures that the model can still be used for rapid detection or real-time monitoring.

Government extranet is an important infrastructure of e-government in China. In the new era, Guangdong Province pro‐ poses to build a new generation government extranet with "resilience, comprehensive integration and extensive coverage", which puts forward higher requirements for the coverage, stability and security of the network. Therefore, combining SPN technology and Super SIM zero trust security gateway technology, this paper discusses the planning, design and construction of a new generation of government extranet.

By analyzing the problems in the traditional teaching mode of C Language Programming， this paper discusses the teach‐ ing framework and implementation plan of applying educational robots to C Language Programming teaching， and discusses the transformation of teacher roles．By introducing educational robots， it is possible to stimulate students' interest in learning， mobilize their internal drive for learning， improve their problem-solving ability， strengthen their teamwork awareness and innovative spirit， and effectively improve teaching effectiveness．

Knowledge graph has been innovated in knowledge representation and management, and is widely used in fields such as intelligent search and question answering, but it has relatively few applications in the field of education and teaching. Based on knowledge graph and natural language processing technology, the main components of the course teaching plan are sorted out, the in‐ formation such as course entities, attributes and relationships are extracted from the talent training scheme and the course teaching plan, and the ontology model of the data science and big data technology professional course system is designed, which realizes the construction of the knowledge graph of the data science and big data technology professional course system.

In view of facial expression recognition technology, the advantages of self-repair network (Self-Cure Network, SCN) fa‐ cial expression recognition algorithm model are used to realize the application of SCN on mobile terminal devices, using the trained model to build facial expression recognition system, and develop a wechat small program as the system platform．After the system test, the results show that the system has high expression recognition accuracy, convenient operation, smooth operation and complete functions．The SCN face expression recognition system based on wechat small program has a wide range of application prospects．

This paper uses the CiteSpace tool and literature statistical methods to conduct a visual knowledge graph analysis of the literature on the transformation of scientific and technological achievements in software quality inspection for the past 10 years. The high-frequency keywords "information security"、"software testing" and "network security" are current research hotspots with strong centrality, and the future will shift to "data encryption" which prominence intensity is 2.71. To promote the transformation of scien‐ tific and technological achievements, third-party quality testing institutions should cooperate deeply with research and development institutions, and leverage the data advantages of market testing and practical capabilities of achievement transformation, and strengthen detailed research on different sub characteristics in the field of quality inspection, and conduct more specific and intuitive analysis of each sub feature at the software application, and establish standards for testing technology through big data analysis. At the same time, it can also rely on artificial intelligence for cross penetration of sub features, achieving the sharing and complementar‐ ity of research and development resources. This paper summarizes the research strength, hotspots and cutting-edge trends of soft‐ ware quality inspection, and provides some reference value for promoting the transformation of subsequent achievements.

With the rapid development of information technology, computer network security has become a focus of social attention. This article explores the teaching innovation and practice of Computer Network Security from the perspective of "ideological and po‐ litical guidance, integration of industry and education", aiming to cultivate outstanding talents with high awareness and skills of net‐ work security. The course utilizes Xuexi Tong APP to carry out blended online and offline teaching. Through the reconstruction of course content, improvement of teaching modes, innovation of teaching methods, integration of ideological and political elements, reform of teaching evaluation, and promotion of teaching effectiveness, the educational philosophy of "moral education first, studentcentered, integration of industry and education, and continuous improvement" is adopted to achieve the deep exploration of ideologi‐ cal and political elements in the course while imparting professional knowledge, integrate professional knowledge of network and se‐ curity with production practice, strengthen school enterprise cooperation, continue teaching reform, and build a learning community for teachers and students.