Voxel51/pheno4d
收藏资源简介:
Pheno4D是一个时空3D点云数据集,包含7株玉米植物(在12天内每天扫描)和7株番茄植物(在20天内每天扫描)。这些植物在温室中盆栽生长,并使用亚毫米精度激光三角测量扫描仪(Perceptron ScanWorks V5安装在ROMER Infinite 2.0测量臂上,σ ≈ 0.012mm)进行测量。每个扫描都是原始、无结构的3D表面点云,覆盖整个盆栽植物,包括土壤,没有颜色/强度通道,只有XYZ几何数据。部分扫描带有手动标注的时间一致性点级实例标签:每片叶子在整个时间序列中保持相同的数字ID,并且玉米数据还提供了两种独立的并行标注方案(叶颈法和叶尖法)。数据集由波恩大学等机构的研究人员创建,由德国研究基金会资助,用于植物表型分析和高级植物分析研究。
Pheno4D is a spatiotemporal 3D point cloud dataset encompassing 7 maize plants (scanned daily over a 12-day period) and 7 tomato plants (scanned daily over a 20-day period). All plants were cultivated in pots within a greenhouse and scanned using a sub-millimeter accuracy laser triangulation scanner (Perceptron ScanWorks V5 mounted on a ROMER Infinite 2.0 measuring arm, with measurement precision σ ≈ 0.012mm). Each scan yields a raw, unstructured 3D surface point cloud covering the entire potted plant including the soil substrate, with no color or intensity channels, only XYZ geometric data. A subset of the scans comes with manually annotated temporally consistent point-level instance labels: each leaf retains a consistent numerical ID across the entire time series, and the maize subset additionally provides two independent parallel annotation frameworks (the "leaf collar method" and the "leaf tip method"). This dataset was developed by researchers from institutions including the University of Bonn, and was funded by the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG) for research on plant phenotyping and advanced plant analysis.




