Skip to main content

Data

The data behind the research.

Point clouds, canopy models and leaf optical measurements from the projects and notes, with their formats documented so they can be read without this site.

4
datasets
3
downloadable here
3
research layers covered
  1. 2024IDigitize

    Potted cotton point cloud from turntable photogrammetry (SfM)

    One growth-chamber cotton plant (sample cotton_20240109-84-5) reconstructed by structure from motion and labelled by organ.

    39,765 points · 3 organ classes · License CC BY-NC 4.0

    Download sfm.bin · 278 KBResearch note
    File format

    Binary, little-endian: Uint16 x, y, z for every point (quantized over the bounding box below), then one Uint8 organ class per point. Classes: 0 main stem, 1 branches and petioles, 2 leaves.

    Bounding box: x 0.250–0.486 · y 0.009–0.412 · z -1.033–-0.718 m

  2. 2025IDigitize

    Cotton point cloud from single-image 3D generative AI

    The same plant reconstructed from a single image with 3D generative AI, for comparison against the SfM reference.

    60,000 points (voxel-downsampled from 177,357) · 3 organ classes · License CC BY-NC 4.0

    Download gen.bin · 420 KBResearch note
    File format

    Binary, little-endian: Uint16 x, y, z for every point (quantized over the bounding box below), then one Uint8 organ class per point. Classes: 0 main stem, 1 branches and petioles, 2 leaves.

    Bounding box: x 0.252–0.493 · y 0.022–0.407 · z -1.016–-0.721 m

  3. 2023IIUnderstandIIIPredict

    Cotton canopy triangle mesh with leaf optical properties

    A 24-plant canopy built from one reconstructed plant, with reflectance and transmittance per optical class, as used for canopy light and photosynthesis modelling.

    1,613,632 triangles in the full model (119,999 in the web file) · leaf area 1.2745 m² · height 0.414 m · License CC BY-NC 4.0

    Download canopy.bin · 1.2 MBResearch note
    File format

    Binary, little-endian: Uint16 x, y, z for the leaf vertices (57,552) then the non-leaf vertices (18,560), quantized over the bounding box; then Uint16 index triples for the leaf triangles, then the non-leaf triangles (indices per group). Optical properties: leaf reflectance 0.1, transmittance 0.05; other organs 0.15 and 0. The web file is simplified per optical class by quadric decimation.

    Bounding box: x -0.475–1.076 · y 0.002–0.411 · z -1.080–-0.224 m

  4. 2025IIUnderstand

    Leaf directional spectra, BRDF parameters and phenotypic traits

    Directional reflectance (400–1000 nm) of maize, rice, cotton and poplar leaves with fitted Cook–Torrance parameters and the measured traits used to predict them.

    270 entries · 4 species · 2 canopy positions · both leaf surfaces

    Available on requestProject pagePaperCode

    As stated in the published article, the data are available from the corresponding author upon reasonable request.

Files downloadable here are licensed CC BY-NC 4.0: free to reuse for non-commercial work with attribution to Liangchao Deng and a link to this page. For commercial use, or to agree on a citable version, get in touch. Contact