On May 28, 2025, a collaborative research paper by students and faculty from the Sino-foreign joint program of the College of International Education (Oxford Brookes College), Chengdu University of Technology, was published in the internationally renowned journal Computers and Electronics in Agriculture (Chinese Academy of Sciences Tier 1, Impact Factor: 7.7). The paper, titled “A novel residual learning of multi-scale feature extraction model for the classification of rice grain varieties”, was co-authored by undergraduates Xudong Li and Yutong Wang of the Sino-foreign cooperative education program in Computer Science and Technology , alongside Nigerian faculty members Dr. Happy Nkanta Monday and Dr. Grace Ugochi Nneji, who served as the corresponding authors.
First page of the published paper
The study addresses the critical agricultural challenge of classification of rice varieties. As a staple food for nearly 50% of the global population, accurately identifying rice varieties plays a vital role in ensuring food security and supporting agricultural productivity. Traditional manual classification methods are inefficient and prone to subjective bias, while existing deep learning models often struggle with robust feature extraction. To tackle these limitations, the research team constructed a large dataset of 75,000 images. Through the use of data augmentation and ensemble learning, the team developed a novel ensemble model that combines a customized attention mechanism with modified residual learning and multi-scale feature extraction. The model achieved near-perfect classification accuracy, reaching 99%, and was further validated using Grad-CAM visualization to confirm the interpretability of its decision-making process. The study not only contributes key technological support to agricultural intelligence, but also shows broad potential for application in grain quality inspection and precision agriculture, offering significant value in ensuring global food security.
Architecture of the ensemble model and Grad-CAM visualization results (from the paper)
The paper was the result of close collaboration between Chinese students and international faculty, highlighting the college’s commitment to fostering “international research collaboration”. Students participated as first authors, engaging in all aspects of the research process, from data collection and model development to manuscript writing. As the first top-tier journal paper produced by the college, this achievement marks a major milestone for the university’s joint education program in interdisciplinary research. Dean Shan Li remarked, “This work exemplifies how high-quality joint education programs can promote deep integration of international teaching and research. It also showcases our students’ creativity in tackling cutting-edge global challenges.”
Publication Details:
Xudong Li, Yutong Wang, Happy Nkanta Monday, Grace Ugochi Nneji. A novel residual learning of multi-scale feature extraction model for the classification of rice grain varieties. Computers and Electronics in Agriculture 237, 110491 (2025).
Paper Link:
https://www.sciencedirect.com/science/article/pii/S0168169925005976?dgcid=coauthor
Computers and Electronics in Agriculture is categorized by the Chinese Academy of Sciences as a Tier 1 journal in both the “Agricultural Engineering” major and minor categories. According to Clarivate’s Web of Science (WOS) classification, it ranks in Q1 (top 25%) within the “Agricultural Engineering” category.
