Related Experiment Video
Updated: Jul 5, 2025
07:10
Isolation of Histone from Sorghum Leaf Tissue for Top Down Mass Spectrometry Profiling of Potential Epigenetic Markers
Published on: March 4, 2021
4.6K
SorghumWeedDataset_Classification and SorghumWeedDataset_Segmentation datasets for classification, detection, and
Michael J Justina1, M Thenmozhi2
1Department of Computer Science and Engineering, School of Computing, SRM Institute of Science and Technology, Kattankulathur Campus, Chennai 603203, India.
Data in Brief
|January 17, 2024
Summary
This study introduces two new public datasets for computer vision research in agriculture, focusing on identifying weeds in sorghum fields. These datasets support the development of precision agriculture techniques for better weed management.
Area of Science:
- Agricultural Science
- Computer Vision
- Machine Learning
Background:
- Weed management is a significant challenge in agriculture, impacting crop yields and requiring labor-intensive methods.
- Traditional agriculture is transitioning towards precision agriculture, necessitating advanced tools for automated crop and weed identification.
- Existing crop-weed datasets, especially from India, are limited for research purposes, hindering the development of computer vision solutions.
Purpose of the Study:
- To develop and release novel, open-access datasets for crop-weed classification and segmentation using computer vision.
- To facilitate research in precision agriculture by providing resources for developing automated weed detection algorithms.
- To address the need for diverse datasets that capture variations in crop spacing and sowing methods.
Main Methods:
- Acquisition of image data from sorghum fields with uniform and random crop spacing.
- Creation of two distinct datasets: 'SorghumWeedDataset_Classification' (4312 samples) and 'SorghumWeedDataset_Segmentation' (5555 annotated segments).
- Manual pixel-wise annotation of segmentation data by experts and verification by agronomists.
Main Results:
- The creation of the first open-access crop-weed datasets from Indian fields for both classification and segmentation tasks.
- Datasets cover diverse agricultural scenarios, including uniform and random crop spacing.
- Public availability of these resources is expected to spur innovation in agricultural computer vision.
Conclusions:
- The released datasets are valuable resources for the research community to advance computer vision applications in agriculture.
- These datasets will aid in developing sophisticated algorithms for precise weed identification and management.
- The availability of these Indian-origin datasets addresses a critical gap in current agricultural research data.
