Back to Search
Start Over
Global Wheat Head Detection 2021: An Improved Dataset for Benchmarking Wheat Head Detection Methods
- Source :
- Plant Phenomics, Vol 2021 (2021)
- Publication Year :
- 2021
- Publisher :
- American Association for the Advancement of Science (AAAS), 2021.
-
Abstract
- The Global Wheat Head Detection (GWHD) dataset was created in 2020 and has assembled 193,634 labelled wheat heads from 4700 RGB images acquired from various acquisition platforms and 7 countries/institutions. With an associated competition hosted in Kaggle, GWHD_2020 has successfully attracted attention from both the computer vision and agricultural science communities. From this first experience, a few avenues for improvements have been identified regarding data size, head diversity, and label reliability. To address these issues, the 2020 dataset has been reexamined, relabeled, and complemented by adding 1722 images from 5 additional countries, allowing for 81,553 additional wheat heads. We now release in 2021 a new version of the Global Wheat Head Detection dataset, which is bigger, more diverse, and less noisy than the GWHD_2020 version.
Details
- Language :
- English
- ISSN :
- 26436515
- Volume :
- 2021
- Database :
- Directory of Open Access Journals
- Journal :
- Plant Phenomics
- Publication Type :
- Academic Journal
- Accession number :
- edsdoj.6363f47b04262b456b55f5e2a3ba9
- Document Type :
- article
- Full Text :
- https://doi.org/10.34133/2021/9846158