This repository contains the instruction for the RoseBlooming dataset. The RoseBloomingdataset allows for stage-specific flower detection. The dataset, consisting of overhead images, contains two rose cultivars and was filmed over a period of months. More details can be found in our paper RoseTracker: A system for automated rose growth monitoring.
- 519 images obtained from actual farms
- 2 classes depend on growth stage: rose_small, rose_large
- Include 2 rose cultivars
- provide in coco format
RoseBlooming dataset Google Drive link.
Non-commercial research purposes only.
├── train
│ ├── train_annotations.coco.json
│ ├── train_{image_id}.jpg
├── valid
│ ├── valid_annotations.coco.json
│ ├── valid_{image_id}.jpg
├── test
├── test_annotations.coco.json
├── test_{image_id}.jpg
- RoseBlooming consist of three files {train / valid / test}
- The .jpg files are the input images, and the .json files contain annotation information.
@article{SHINODA2023100271,
title = {RoseTracker: A system for automated rose growth monitoring},
journal = {Smart Agricultural Technology},
volume = {5},
pages = {100271},
year = {2023},
issn = {2772-3755},
doi = {https://doi.org/10.1016/j.atech.2023.100271},
url = {https://www.sciencedirect.com/science/article/pii/S2772375523001004},
author = {Risa Shinoda and Ko Motoki and Kensho Hara and Hirokatsu Kataoka and Ryohei Nakano and Tetsuya Nakazaki and Ryozo Noguchi},
}