PlantDepth: A Benchmark Dataset for Plant Monocular Depth Estimation and 3D Reconstruction

PlantDepth is a benchmark dataset for plant monocular depth estimation and 3D reconstruction, associated with the PlantMDE project and the work titled 3D Plant Phenotyping from a Single Image: Generalizable Plant Monocular Depth Estimation via Dataset Mixing. The dataset provides rendered RGB-D plant images derived from publicly available plant 3D datasets. The archive contains eight sub-datasets: Crops3D, ETH_BeetRoot, GScatter, PLANest, Plant3D, PlantStereo, Soybeanmvs, and WUR_DwarfTomato. Dataset loading code and usage examples are available in the PlantMDE repository. This deposit distributes the original 14,611,995,103-byte PlantDepth.zip as 30 binary parts, named PlantDepth.zip.part-000 through PlantDepth.zip.part-029, each up to 500,000,000 bytes. It also includes SHA256SUMS and README_DOWNLOAD.txt. Download all parts and the checksum file into one directory, concatenate the parts in numerical order, verify SHA-256 checksums, and extract the reconstructed ZIP. Individual parts cannot be extracted separately. cat PlantDepth.zip.part-[0-9][0-9][0-9] > PlantDepth.zip && sha256sum -c SHA256SUMS && unzip PlantDepth.zip On macOS, use shasum -a 256 -c SHA256SUMS for verification. See README_DOWNLOAD.txt for further instructions.