To put into practice the educational philosophy of “enlightening minds through culture, nourishing hearts through culture, promoting learning through culture and cultivating people through culture”, stimulate the innovative vitality of postgraduate students and promote postgraduate academic exchange, the 18th “Shangxue Forum” postgraduate academic report session of 2026, jointly organised by the Graduate School, the Science and Technology Office and the College of Electronic Information and Automation, was successfully held recently. The session built a bridge of academic exchange for sharing cutting-edge research results. Professor Lin Liyuan, a discipline leader of the College, was invited to share her research journey, and three master’s students of the College were invited to report on their research results, exploring cutting-edge achievements and discussing research growth with fellow students.

Professor Lin Liyuan gave her sharing under the title “From Real-World Needs to Methodological Innovation — Research Practice in Non-Contact Human Perception”. Drawing on her long experience of working at the research frontline, she started from real application needs and described the complete path of research practice, from identifying research topics to methodological innovation. She encouraged the young postgraduate students present to extract scientific problems from real-world scenarios, to be unafraid of repeated trial and error in experiments, and to refine innovative ideas through sustained effort, providing valuable guidance for the students’ research planning.

▲ Professor Lin Liyuan delivering her presentation

Liu Zecheng delivered a themed report entitled “Research on AI Vision-Based Posture Recognition Algorithms”. In his presentation, he elaborated on the model architecture design and the experimental validation process, analysed the strengths of the current method and its remaining difficulties, and offered new ideas for research on behaviour recognition. He also shared his own experiences and reflections on building datasets and tuning model parameters, demonstrating the perseverance and thinking of a young researcher exploring his research topic.

▲ Liu Zecheng presenting his research

Li Zheng presented his research results under the title “Research on Vision-Language Model-Based Anomaly Detection Methods for Cross-Domain Scenarios”. Addressing pain points in industrial and complex scenarios, such as the difficulty of cross-domain transfer and insufficient detection accuracy, he introduced a research scheme for anomaly detection based on vision-language models, analysed the challenges brought by feature shift in cross-domain scenarios, and explained key technologies such as model feature alignment and optimisation strategies, illustrating the application value of artificial intelligence technology in anomaly monitoring scenarios.

Shi Kaiyue reported on her research entitled “Ultra-Wide-Field Fundus Image Diagnosis Based on Visual Semantic Refinement Feature Learning”. She combined computer vision technology with medical imaging diagnosis, using visual semantic refinement feature learning to assist the screening and diagnosis of fundus diseases for ultra-wide-field fundus images. She shared her reflections on interdisciplinary research and her thinking on balancing algorithm performance with actual clinical needs in medical image research, letting the teachers and students present feel the power of research to serve people’s livelihood.

▲ Postgraduate students presenting their research

Going forward, the university will continue to rely on the “Shangxue Forum” series of activities to make academic exchange a regular practice, encourage more postgraduate students to devote themselves to scientific exploration, gain new insights through exchange and mutual learning, and bravely climb the peaks of research in overcoming difficulties.