Research on Improving YOLOv11n for Siraitia grosvenorii Pistil Detection Using SCConv and CoordAtt Dual-Module Synergy
Year: 2026
Type: article
Abstract: Precision-targeted pollination places strict demands on flower organ detection in field environments. In field conditions, pistil detection in Siraitia grosvenorii remains difficult because the targets are small, often occluded, and require accurate localization under lightweight model constraints. To address these challenges, we develop an improved YOLOv11n-based method for Siraitia grosvenorii pistil detection in precision pollination tasks. The model incorporates SCConv, CoordAtt, and SimAM to improve feature extraction and foreground discrimination for small pistil targets in complex backgrounds. The main contribution of this work lies in task-oriented module integration and lightweight optimization for tiny pistil detection, rather than in proposing a new generic detection operator. Experiments on the self-built dataset show that the improved model achieves 82.17% [email protected], 40.38% [email protected]:0.95, and 86.20% precision, improving upon the YOLOv11n baseline by 2.92, 2.28, and 9.84 percentage points, respectively. Recall decreases from 78.46% to 76.50%, suggesting a precision-oriented trade-off in the current setting. With only 2.89 M parameters and 7.22 GFLOPs, the model maintains a lightweight architecture while achieving improved detection performance for targeted pollination tasks. These findings support the feasibility of the proposed method for Siraitia grosvenorii pistil detection in intelligent pollination applications.
Source: Applied Sciences
DOI: 10.3390/app16084057
Language: English
Topic: Smart Agriculture and AI
Subfield: Plant Science
Domain: Life Sciences
SDG: Zero hunger
Open Access status: gold
Estimated APC paid: $2,896
Award: G5131997164
OpenAlex ID: W7155070844