ORIGINAL RESEARCH article
Front. Plant Sci.
Sec. Technical Advances in Plant Science
ECS-Tea: A Bio-Inspired High-Precision Detection and Localization Algorithm for Young Shoots of Pu-erh Tea
Provisionally accepted
- Southwest Forestry University, Kunming, 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
Pu-erh tea, celebrated for its unique ecological significance and economic worth, necessitates precise and efficient bud harvesting to advance intelligent agricultural operations. To this end, we propose ECS-Tea, a bio-inspired and lightweight detection–localization framework built upon YOLOv11-Pose, tailored for accurate bud recognition and keypoint estimation in complex natural environments. This framework incorporates four fundamental modules:(1) A lightweight EfficientNetV2 backbone designed for highly efficient feature representation and encoding;(2) A Cross-Scale Feature Fusion (CSFF) module that strengthens the representation of multi-scale contextual information;(3) A Spatial–Channel Synergistic Attention (SCSA) mechanism enabling fine-grained and discriminative keypoint feature modeling;(4) An adaptive multi-frame depth fusion strategy designed to enhance the precision and robustness of 3D localization. The ECS-Tea framework was trained and validated on a dedicated dataset specifically curated for Pu-erh tea bud detection. Experimental results reveal that the proposed model attains 98.7% target detection accuracy and 95.3% keypoint detection accuracy, while maintaining a compact architecture of 3.3 MB, computational cost of 4.5 GFLOPs, and an impressive inference speed of 370.4 FPS.Relative to the baseline YOLOv11-Pose, ECS-Tea yields a marked enhancement in keypoint detection, with mAP@0.5(K) rising by 4.9%, recall R(K) by 3.8%, and precision P(K) by 3.4%, while sustaining or marginally improving object detection metrics. These findings underscore the model’s capability to strike an effective balance between accuracy and computational efficiency, while empirically confirming the complementary contributions of its integrated modules. Consequently, ECS-Tea offers a robust, real-time, and deployable solution for high-precision tea bud harvesting in unstructured field environments, bridging the gap between algorithmic sophistication and practical agricultural application.
Keywords: Pu-erh tea, YOLOPose, object detection, Pose estimation, Depth camera, smartagriculture
Received: 02 Sep 2025; Accepted: 30 Oct 2025.
Copyright: © 2025 Wang, Li, Xu, Ti, Jiang, Liao, Li and Li. 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: Wei Li, liwei@swfu.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.