Xiaojun Jin received his B.S., M.S., and Ph.D. degrees in Mechanical Engineering from Nanjing Forestry University (ranked 72nd in China) in 2009, 2012, and 2024, respectively. He is currently an Associate Professor at the Peking University Institute of Advanced Agricultural Sciences, where his research focuses on robotics and sensing technologies for smart agriculture, with a particular emphasis on precision weed control. Before joining PKU-IAAS, Dr. Jin worked as a Principal Engineer at SAIC Mobility Co., Ltd. from 2019 to 2023, developing software for SAIC Motor’s mobility platform. Prior to that, he served as a Senior Software Engineer at ArcSoft, Inc., specializing in retail and OEM software design, with a focus on architecture, modular design, continuous integration, and project management. Dr. Jin has published over 40 peer-reviewed papers in reputable journals and holds 24 patents. Additionally, he has registered 24 software copyrights. His current research interests include machine vision, artificial intelligence, and robotics. For more details, please download Dr. Jin’s resumé.
Weed coverage percentage was classified into four levels (25%, 50%, 75%, and 100%) based on the number of weed coverage estimation cells containing weeds.
The grid cells were marked as spraying areas if the inference result indicated that they contained weeds. Only those nozzles corresponding to those cells infested with weeds were turned on, thus realizing a smart sensing and spraying system.
A deep-learning model was used to detect vegetables and draw bounding boxes around them. Thereafter, the plants falling out of the bounding boxes were considered as weeds. This strategy avoids dealing with various weed species and thus significantly reduces the overall complexity of weed detection in vegetable fields.