Datasets:

Project-AgML
/

papple_rgbd_amodal_segmentation

depth unknown	objects dict	split stringclasses 1 value
"k05VTVBZAQBGAHsnZGVzY3InOiAnPGY4JywgJ2ZvcnRyYW5fb3JkZXInOiBUcnVlLCAnc2hhcGUnOiAoMTMwMCwgMTMwMCksIH0(...TRUNCATED)	{"bbox":[[139.0,1118.0,46.0,42.0]],"categories":[0],"apple_id":[1],"polygon":[[165.0,1118.0,164.0,11(...TRUNCATED)	train
"k05VTVBZAQBGAHsnZGVzY3InOiAnPGY4JywgJ2ZvcnRyYW5fb3JkZXInOiBUcnVlLCAnc2hhcGUnOiAoMTMwMCwgMTMwMCksIH0(...TRUNCATED)	{"bbox":[[1111.0,272.0,49.0,211.0],[937.0,321.0,194.0,195.0],[1079.0,932.0,66.0,91.0],[1011.0,1055.0(...TRUNCATED)	train
"k05VTVBZAQBGAHsnZGVzY3InOiAnPGY4JywgJ2ZvcnRyYW5fb3JkZXInOiBUcnVlLCAnc2hhcGUnOiAoMTMwMCwgMTMwMCksIH0(...TRUNCATED)	{"bbox":[[1132.0,220.0,28.0,124.0],[278.0,269.0,182.0,217.0],[139.0,320.0,157.0,197.0],[582.0,886.0,(...TRUNCATED)	train
"k05VTVBZAQBGAHsnZGVzY3InOiAnPGY4JywgJ2ZvcnRyYW5fb3JkZXInOiBUcnVlLCAnc2hhcGUnOiAoMTMwMCwgMTMwMCksIH0(...TRUNCATED)	{"bbox":[[139.0,308.0,51.0,149.0]],"categories":[0],"apple_id":[1],"polygon":[[163.0,308.0,162.0,309(...TRUNCATED)	train
"k05VTVBZAQBGAHsnZGVzY3InOiAnPGY4JywgJ2ZvcnRyYW5fb3JkZXInOiBUcnVlLCAnc2hhcGUnOiAoMTMwMCwgMTMwMCksIH0(...TRUNCATED)	{"bbox":[[1081.0,141.0,60.0,67.0],[950.0,243.0,119.0,201.0]],"categories":[0,0],"apple_id":[19,20],"(...TRUNCATED)	train
"k05VTVBZAQBGAHsnZGVzY3InOiAnPGY4JywgJ2ZvcnRyYW5fb3JkZXInOiBUcnVlLCAnc2hhcGUnOiAoMTMwMCwgMTMwMCksIH0(...TRUNCATED)	{"bbox":[[621.0,245.0,237.0,263.0]],"categories":[0],"apple_id":[7],"polygon":[[722.0,245.0,721.0,24(...TRUNCATED)	train
"k05VTVBZAQBGAHsnZGVzY3InOiAnPGY4JywgJ2ZvcnRyYW5fb3JkZXInOiBUcnVlLCAnc2hhcGUnOiAoMTMwMCwgMTMwMCksIH0(...TRUNCATED)	{"bbox":[[373.0,1031.0,175.0,126.0],[211.0,1098.0,173.0,62.0],[806.0,371.0,88.0,88.0],[488.0,827.0,1(...TRUNCATED)	train
"k05VTVBZAQBGAHsnZGVzY3InOiAnPGY4JywgJ2ZvcnRyYW5fb3JkZXInOiBUcnVlLCAnc2hhcGUnOiAoMTMwMCwgMTMwMCksIH0(...TRUNCATED)	{"bbox":[[888.0,1018.0,186.0,139.0],[386.0,969.0,160.0,149.0],[564.0,751.0,184.0,120.0],[859.0,611.0(...TRUNCATED)	train
"k05VTVBZAQBGAHsnZGVzY3InOiAnPGY4JywgJ2ZvcnRyYW5fb3JkZXInOiBUcnVlLCAnc2hhcGUnOiAoMTMwMCwgMTMwMCksIH0(...TRUNCATED)	{"bbox":[[141.0,1006.0,102.0,151.0]],"categories":[0],"apple_id":[1],"polygon":[[143.0,1006.0,142.0,(...TRUNCATED)	train
"k05VTVBZAQBGAHsnZGVzY3InOiAnPGY4JywgJ2ZvcnRyYW5fb3JkZXInOiBUcnVlLCAnc2hhcGUnOiAoMTMwMCwgMTMwMCksIH0(...TRUNCATED)	{"bbox":[[1033.0,289.0,127.0,195.0],[1091.0,995.0,68.0,161.0],[807.0,604.0,186.0,220.0],[527.0,149.0(...TRUNCATED)	train

End of preview. Expand in Data Studio

Papple Rgbd Amodal Segmentation

PApple_RGB-D-Size: 3,925 RGB-D images (1300x1300) of Fuji apple trees at growth stages BBCH77 and BBCH85, released with a study on estimating the size of on-tree fruit hidden behind occlusions. Each apple carries a modal (visible) outline and an amodal (complete) outline, with matching boxes and its apple ID. The mask column marks the visible apple pixels. Depth maps produced by photogrammetry are kept in their own column as the original float64 .npy arrays, with the source's train/val/test split; the record gives a focal length of 5805.34 px for every image. The dataset contains 3,925 images with 15,335 bounding box annotations across 1 category.

This dataset is indexed on https://project-agml.github.io/ as part of the AgML python library.

The original train/test/val split has been preserved in the split column.

Citation

@article{genmola2023looking,
  title={Looking behind occlusions: A study on amodal segmentation for robust on-tree apple fruit size estimation},
  author={Gené-Mola, Jordi and Ferrer-Ferrer, Mar and Gregorio, Eduard and Blok, Pieter M. and Hemming, Jochen and Morros, Josep-Ramon and Rosell-Polo, Joan R. and Vilaplana, Verónica and Ruiz-Hidalgo, Javier},
  journal={Computers and Electronics in Agriculture},
  volume={209},
  pages={107854},
  year={2023},
  publisher={Elsevier}
}

The dataset itself can be cited as:

Mola, J. G., Ferrer, M. F., López, E. G., Polo, J. R. R., Vilaplana, V., Hidalgo, J. R., & Morros, J. R. (2022). PApple_RGB-D-Size dataset [Dataset]. Zenodo. https://doi.org/10.5281/ZENODO.7260694

This dataset was reformatted from its original format to match HuggingFace standards.

Downloads last month: 79