Dataset for Detecting Small Onions and Flowering Plants

Jenn Hoskins
25th July, 2024

Dataset for Detecting Small Onions and Flowering Plants

Image Source: Natural Science News, 2024

Key Findings

  • Researchers at Universidad EAFIT created a digital image dataset to detect green onion and foliage flower crops in Medellín, Colombia
  • The dataset includes 245 images labeled with green onion, foliage flowers, and non-crop areas, with 4315 instances annotated for machine learning
  • This dataset aims to improve crop detection accuracy and efficiency, reducing the need for manual fieldwork
Precision Agriculture (PA) aims to optimize agricultural practices through the use of technology. One of the critical areas of research in PA is the detection of different crop types using aerial images. This capability is essential for establishing crop inventories, determining planting areas, and forecasting crop yields, which are crucial for food markets and public entities that support small farmers. A recent study by researchers at Universidad EAFIT addresses the need for public access to a digital image dataset specifically for detecting green onion and foliage flower crops in the rural area of Medellín City, Colombia[1]. The dataset introduced in this study consists of 245 images labeled with three categories: green onion (Allium fistulosum), foliage flowers (Solidago Canadensis and Aster divaricatus), and non-crop areas prepared for planting. A total of 4315 instances were annotated using the polygon method, which allows for the training of machine learning algorithms for detection using bounding boxes or segmentation in the COCO format. The significance of this dataset lies in its potential to improve crop detection accuracy and efficiency. Traditional methods for identifying crop types often rely on manual fieldwork, which can be time-consuming and prone to error. The introduction of high-resolution digital images and advanced machine learning techniques offers a more reliable and scalable solution. This study builds on earlier research that demonstrated the effectiveness of deep learning models in agricultural applications. For instance, a study conducted in Southern Bangladesh evaluated three different full convolutional neural network (F-CNN) models to detect functional field boundaries from satellite imagery[2]. Among the models tested, DenseNet achieved the highest precision, recall, and F-1 score, indicating its potential for accurately identifying small, irregularly shaped agricultural fields. This finding underscores the value of using advanced neural architectures for crop detection tasks. Furthermore, the importance of field data for adequate agricultural monitoring has been highlighted in previous research. A dataset from the west of Bahia state, Brazil, provided monthly information on 16 land use classes for 1854 fields over an agricultural year[3]. This dataset was instrumental in developing new pattern recognition methods for agricultural land use mapping and monitoring. The integration of such datasets can enhance the accuracy of remote sensing applications in agriculture. The current study by Universidad EAFIT contributes to this growing body of knowledge by providing a specialized dataset for green onion and foliage flower crops. The availability of this dataset enables researchers to train and validate machine learning models tailored to these specific crops, potentially improving the precision of crop detection in similar agricultural settings. Moreover, the use of digital image datasets aligns with the broader trend of employing UAVs (Unmanned Aerial Vehicles) and deep learning algorithms for precision agriculture. UAVs offer a non-invasive and efficient means of collecting high-resolution images, which can be analyzed to detect various agricultural issues, including plant diseases and pests[4]. The integration of UAV technology with the dataset from Universidad EAFIT could further enhance the capabilities of precision agriculture by providing timely and accurate crop monitoring. In summary, the introduction of a publicly accessible digital image dataset for detecting green onion and foliage flower crops by Universidad EAFIT represents a significant advancement in precision agriculture. By leveraging high-resolution images and advanced machine learning techniques, this dataset has the potential to improve crop detection accuracy and efficiency, thereby supporting small farmers and contributing to food security. This study, along with previous research, underscores the importance of integrating advanced technologies and comprehensive datasets to address the challenges faced by modern agriculture.

AgricultureBiochemPlant Science

References

Main Study

1) OnionFoliageSET: Labeled dataset for small onion and foliage flower crop detection.

Published 24th July, 2024

https://doi.org/10.1016/j.dib.2024.110679

Related Studies

2) Detecting functional field units from satellite images in smallholder farming systems using a deep learning based computer vision approach: A case study from Bangladesh.

https://doi.org/10.1016/j.rsase.2020.100413

3) LEM+ dataset: For agricultural remote sensing applications.

https://doi.org/10.1016/j.dib.2020.106553

4) A survey on deep learning-based identification of plant and crop diseases from UAV-based aerial images.

https://doi.org/10.1007/s10586-022-03627-x


Related Articles

An unhandled error has occurred. Reload 🗙