提升果园中沙田柚检测精度，应对光照、遮挡等实际挑战
A Multi-Strategy Framework for Enhancing Shatian Pomelo Detection in Real-World Orchards
- 设计多策略框架，融合全局可见性卷积与动态特征增强
- 在真实果园数据集上实现84.3% [email protected]，显著优于现有模型
- 适合农业视觉检测、作物智能管理领域的研究与应用
沙田柚的果园检测对产量估算和精益生产至关重要，但针对理想数据集训练的模型在实际场景中常因设备相关的色调偏移、光照变化、尺度差异大及频繁遮挡而性能下降。本文构建了包含真实果园图像与精选网络图像的多场景数据集STP-AgriData，并通过对比度/亮度增强模拟不稳定的光照条件。为更好应对尺度变化与遮挡问题，提出REAS-Det，包含全局-选择性可见性卷积（GSV-Conv），在全局语义引导下扩展可见特征空间，同时保持高效的空间聚合；此外引入C3RFEM、MultiSEAM与Soft-NMS实现精细分离与定位。在STP-AgriData上，REAS-Det达到86.5%精确率、77.2%召回率、84.3% [email protected]，以及53.6% [email protected]:0.95，显著优于近期检测器，在真实果园环境中表现更鲁棒。源代码已开源：https://github.com/Genk641/REAS-Det。
原文摘要 · Abstract (English)
Shatian pomelo detection in orchards is essential for yield estimation and lean production, but models tuned to ideal datasets often degrade in practice due to device-dependent tone shifts, illumination changes, large scale variation, and frequent occlusion. We introduce STP-AgriData, a multi-scenario dataset combining real-orchard imagery with curated web images, and apply contrast/brightness augmentations to emulate unstable lighting. To better address scale and occlusion, we propose REAS-Det, featuring Global-Selective Visibility Convolution (GSV-Conv) that expands the visible feature space under global semantic guidance while retaining efficient spatial aggregation, plus C3RFEM, MultiSEAM, and Soft-NMS for refined separation and localization. On STP-AgriData, REAS-Det achieves 86.5% precision, 77.2% recall, 84.3% [email protected], and 53.6% [email protected]:0.95, outperforming recent detectors and improving robustness in real orchard environments. The source code is available at: https://github.com/Genk641/REAS-Det.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。