Intelligent Vision Technology: Conduct research on intelligent visual perception and understanding, and achieve multiple results in aspects such as anti-interference intelligent perception, adaptive learning models, and lightweight computing. These results serve fields such as agriculture, transportation, and border security.Utilize deep neural network technology to carry out research on image quality evaluation and image aesthetic evaluation, and obtain multiple intelligent image evaluation technologies, which can be applied to scenarios such as photography, selection of news pictures, artistic creation, and evaluation.Employ theoretical technologies such as encoder-decoder architecture, visual attention mechanism, and knowledge reasoning to conduct research on image description, and obtain multiple new methods, which can be used in fields such as advertising design, news reporting, and assisted education.Starting from multimodal perception, combining technologies such as attention mechanism, knowledge mining, and causal reasoning to conduct research on visual question answering. The research results can be applied to fields such as assistance for the visually impaired and human-computer interaction.Aiming at the requirements of high-precision industrial inspection, independently develop software and hardware such as intelligent cable measuring instruments, high-performance flash measurement software, and mobile stitching large-size measurement software. Many technical indicators are internationally leading, and these can be applied to indu Faculty members of the school have been awarded 2 National Science and Technology Progress Awards, 1 National Defense Science and Technology Award, and 10 provincial/ministerial-level science and technology awards. They have led multiple national-level research initiatives, including projects under the National High-Tech R&D Program (863 Program), the National Basic Research Program (973 Program), the National Key R&D Program, the Australian Research Council (ARC) Projects, and Key Projects under the National Natural Science Foundation of China (NSFC) Joint Funds. stries such as construction, 3C electronics, and automobiles.Develop a low-cost crop growth monitoring and phenotypic analysis system to serve the cultivation and management of crops such as lettuce and tomatoes.Develop educational software such as the track and field sports video analysis system and the intelligent evaluation system for normal students' teaching skills, providing intelligent vision solutions for the education industry.

Artificial Intelligence + Biomedicine: The Artificial Intelligence Research Institute of Guangxi Normal University focuses on the urgent needs of the national pharmaceutical innovation and development strategy and the development of Guangxi's pharmaceutical and health industry. Relying on the first-level discipline doctoral program in software engineering, the professional doctoral program in biology and medicine, as well as provincial and ministerial-level scientific research platforms such as the Key Laboratory of Medicinal Resource Chemistry and Drug Molecular Engineering of the Ministry of Education, the Guangxi Key Laboratory of Multi-source Information Mining and Security, the Key Laboratory of Stem Cells and Medical Biotechnology in Guangxi Universities, and the Key Laboratory of Data Science Interdisciplinary Research in Guangxi Universities, the institute focuses on the innovative applications of technologies such as machine learning, deep learning, and intelligent optimization algorithms in the fields of precision medicine, drug research and development, and biomedical literature analysis. It conducts research on flow cytometry analysis, identification of cancer driver pathways, research and development of Guangxi's characteristic ethnic medicines, and research and development of biomedical large models.
 In recent years, it has carried out research on the development and utilization of important medicinal resources in Guangxi in cooperation with well-known domestic traditional Chinese medicine enterprises such as Guilin Sanjin Pharmaceutical Co., Ltd., Guilin China Resources Tianhe Pharmaceutical Co., Ltd., and Guilin Pharma Co., Ltd. It has carried out 34 horizontal cooperation projects, and 18 achievement technologies have been transferred, providing important scientific and technological support for Guangxi's pharmaceutical manufacturing industry worth hundreds of billions of yuan. The developed high-performance flow cytometry analysis instruments and systems are exported to all over the world and have currently been promoted in more than 180 countries.

Intelligent Education: Through the deep integration of educational knowledge graphs, large models, and personalized applications, it promotes a paradigm breakthrough and scenario reconstruction in educational intelligence. With the construction of multimodal educational knowledge graphs as the core feature, by modeling subject knowledge ontology and semantically aligning cross-modal data (text, video, test questions, etc.), a dynamically evolving educational cognitive network is constructed to solve the problems of fragmentation and insufficient correlation of traditional educational resources.At the technical level, the research team proposes a vertical domain enhancement framework for educational large models, breaking through the bottlenecks of general large models such as insufficient logical rigor and poor interpretability of reasoning in educational scenarios. Through the integration of the structured constraints of subject knowledge graphs and the training of educational behavior data, large models can deeply understand subject semantics (such as mathematical symbol logic and physical causal reasoning) and generate intelligent question answering and problem-solving strategies that meet teaching objectives.
 At the application level, it focuses on the dynamic closed-loop optimization of personalized learning. Based on multi-source learning situation data (classroom interaction, homework performance, emotional feedback), a digital portrait of learners is constructed, and the real-time iteration of diagnosis-recommendation-feedback is realized by combining reinforcement learning algorithms.The research team further explores the protection and utilization of educational intellectual property rights. By combining domestic and foreign patent datasets and using large model and knowledge graph technologies, it deeply mines the associations and technological evolution among patents, providing accurate and efficient authorization and value evaluation support for universities. This technology can not only identify the technological potential and market value of patents but also conduct cross-domain analysis and evaluation from a global perspective in combination with international dynamics. Through the construction of knowledge graphs and deep reasoning of large models, the efficiency and accuracy of patent authorization decisions are significantly improved, promoting the effective transformation of university patent technologies, providing strong support for the industrial application of intellectual property rights, and thus accelerating the release and in-depth development of patent value.

Intelligent Education: Through the deep integration of educational knowledge graphs, large models, and personalized applications, it promotes a paradigm breakthrough and scenario reconstruction in educational intelligence. With the construction of multimodal educational knowledge graphs as the core feature, by conducting subject knowledge ontology modeling and semantic alignment of cross-modal data (text, videos, test questions, etc.), a dynamically evolving educational cognitive network is constructed to address the issues of fragmentation and insufficient correlation of traditional educational resources.At the technical level, the research team has proposed a vertical domain enhancement framework for educational large models, breaking through the bottlenecks such as insufficient logical rigor and poor interpretability of reasoning of general large models in educational scenarios. By integrating the structured constraints of subject knowledge graphs and the training of educational behavior data, large models can deeply understand subject semantics (such as mathematical symbol logic and physical causal reasoning), and generate intelligent question answering and problem-solving strategies that are in line with teaching objectives.At the application level, it focuses on the dynamic closed-loop optimization of personalized learning. Based on multi-source learning situation data (classroom interactions, homework performance, emotional feedback), a digital portrait of learners is constructed, and the real-time iteration of diagnosis-recommendation-feedback is achieved by combining reinforcement learning algorithms.The research team further explores the protection and utilization of educational intellectual property rights. By integrating domestic and foreign patent datasets and applying large model and knowledge graph technologies, it deeply excavates the associations among patents and technological evolution, providing accurate and efficient authorization and value assessment support for universities. This technology can not only identify the technological potential and market value of patents, but also conduct cross-domain analysis and assessment from a global perspective in combination with international developments. Through the construction of knowledge graphs and deep reasoning of large models, the efficiency and accuracy of patent authorization decisions are significantly improved, promoting the effective transformation of university patent technologies, providing strong support for the industrial application of intellectual property rights, and further accelerating the release and in-depth development of patent value.