Answers are grounded in this paper and cite their sources.
Manual monitoring of plastic film and drip tape laying quality during mulch-covered seed drill operations is limited by high labor intensity, delayed fault detection, and unstable accuracy. To solve these problems, this study proposes a multi-source data acquisition method combining depth cameras and inductive proximity switches, and develops a dedicated monitoring system with a Raspberry Pi 5 as the core controller. Aiming at the key challenges of accurate mulch film segmentation and limited computing power of edge devices, an enhanced lightweight YOLOv11n-seg-DBM segmentation model is proposed, it integrates a C3k2-DWR module to enhance multi-scale feature extraction, a C2BRA module to optimize cross-layer feature interaction, an MFM module for dynamic feature fusion, and a Focaler-CIoU loss function to improve hard sample mining. After pruning and quantization, the model is successfully deployed on edge devices. The system achieves quantitative detection of mulch exposed area and soil coverage width via sub-pixel edge detection and depth information calculation, and realizes real-time fault diagnosis and alarm for film and tape breakage through proximity switch signal sequence analysis. Ablation test results show the improved model outperforms the baseline by 3.52 percentage points in mIoU and 3.35 percentage points in MPA, with parameters reduced to 2.75M Field trials confirm average monitoring accuracies of 93.05% for mulch exposed area and 93.02% for soil coverage width, with average response times of 1.21 s for film breakage and 1.59 s for tape breakage. This study provides an effective technical solution for intelligent operation monitoring and quality improvement of mulch-covered seed drills.
No takes yet. Share an insight, caveat, or question.
Sun et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: