Kefeng Li

Macao Polytechnic University, Shanghai Polytechnic University

Papers

Total Citations

H-Index

About

Kefeng Li is a researcher advancing the frontiers of efficient deep learning and computer vision, with a focus on deploying intelligent systems on resource-constrained devices. His primary research areas include lightweight convolutional neural networks (CNNs), object detection, and environmental monitoring through computer vision. Li’s major contributions center on developing novel model compression techniques, such as fusion pruning, which dramatically reduce computational overhead without sacrificing accuracy. His work on the "Novel Fusion Pruning-Processed Lightweight CNN" (2024, 11 citations) addresses the critical challenge of running complex AI models on smartphones and consumer electronics, enabling real-time local object recognition. Furthermore, his "YOLOv8-RepGhostEMA" model (2024, 5 citations) tackles the pressing issue of underwater trash detection, combining efficient architecture design with attention mechanisms to aid autonomous underwater robots in environmental cleanup. These contributions demonstrate Li’s commitment to bridging the gap between theoretical AI advances and practical, deployable solutions. His research is particularly notable for its direct impact on sustainability and smart device functionality, making him a key figure in the push toward efficient, real-world AI applications.