ORIGINAL RESEARCH article
Front. Plant Sci.
Sec. Sustainable and Intelligent Phytoprotection
This article is part of the Research TopicAI-Driven Plant Intelligence: Bridging Multimodal Sensing, Adaptive Learning, and Ecological Sustainability in Precision Plant ProtectionView all 9 articles
SEAFEC: A Spatial–Edge Adaptive Convolution for Multi-scale and Boundary-aware Plant Disease and Weed Imagery
Provisionally accepted
- Guangxi University, Nanning, China
Select one of your emails
You have multiple emails registered with Frontiers:
Notify me on publication
Please enter your email address:
If you already have an account, please login
You don't have a Frontiers account ? You can register here
Plant diseases and weeds are among the leading biological threats to global crop pro-duction, causing substantial yield losses if not accurately identified and managed in time. While deep learning has advanced the automated analysis of agricultural imagery, existing approaches often fail under two persistent challenges: large multi-scale variations of disease lesions and weeds, and blurred or indistinct boundaries between healthy and infected tissues or between crops and weeds. These issues reduce the reliability of recognition in real-world agricultural fields. To address this, we propose SEAFEC (Spatial–Edge Adaptive Feature Enhancement Convolution), a novel convolutional module that jointly enhances scale adaptivity and boundary precision. SEA-FEC employs a dual-branch design: one branch dynamically adjusts receptive fields for targets of varying scales, while the other explicitly strengthens edge features at multiple scales, supported by channel interaction to unify structural and semantic cues. Across three representative tasks—plant disease classification, corn leaf disease detection, and sugarcane–weed segmentation—SEAFEC achieved consistent improvements (+1.8% accuracy, +2.5% mAP, +3.4% mIoU), with notable gains in boundary-sensitive cases. These results highlight SEAFEC as a general-purpose enhancement module, applicable across different recognition paradigms in agricultural vision. These results demonstrate that SEAFEC provides a unified and efficient solution for tackling the scale–boundary challenges of agricultural imagery, supporting more reliable disease diagnosis and precision weed management in sustainable crop production.
Keywords: crop protection, Convolutional neural networks (CNNs), Boundary-aware featuremodeling, smart agriculture, Multi-scale representation data structure
Received: 29 Aug 2025; Accepted: 21 Nov 2025.
Copyright: © 2025 Sun, Liu, He and Huang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
* Correspondence: Cuimin Sun, cmsun@gxu.edu.cn
Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.