Productive Crop Fields Dataset

Citation Author(s):
Eduardo
Garcia do Nascimento
John
Just
Jurandy
Almeida
Tiago
Almeida
Submitted by:
Eduardo Garcia ...
Last updated:
Fri, 07/14/2023 - 19:15
DOI:
10.21227/rr8k-5r07
Data Format:
Research Article Link:
License:
0
0 ratings - Please login to submit your rating.

Abstract 

In precision agriculture, detecting productive crop fields is an essential practice that allows the farmer to evaluate operating performance separately and compare different seed varieties, pesticides, and fertilizers. However, manually identifying productive fields is often time-consuming, costly, and subjective. Previous studies explore different methods to detect crop fields using advanced machine learning algorithms to support the specialists’ decisions, but they often lack good quality labeled data. In this context, we propose a high-quality dataset generated by machine operation combined with Sentinel-2 images tracked over time. As far as we know, it is the first one to overcome the lack of labeled samples by using this technique. In sequence, we apply a semi-supervised classification of unlabeled data and state-of-the-art supervised and self-supervised deep learning methods to detect productive crop fields automatically. Finally, the results demonstrate high accuracy in Positive Unlabeled learning, which perfectly fits the problem where we have high confidence in the positive samples. Best performances have been found in Triplet Loss Siamese given the existence of an accurate dataset and Contrastive Learning considering situations where we do not have a comprehensive labeled dataset available.

Instructions: 

The dataset used for this project consists of satellite imagery samples containing labeled crop field data. The dataset includes various satellite bands, such as near-infrared (NIR), red, green, and blue, which provide valuable information about the vegetation and presence of crops. The labeled data indicates the presence or absence of productive crop fields in the images. The dataset is indexed by L12 hexes supported by H3 library developed by Uber. The hexes are processed to create timeseries samples containing various image dates to cover most of the year and the diverse weather conditions as well as cloud covering. The representation of a timeseries sample where each row is an image date and each column a Sentinel-2 band is represented below.

For more information, please access the following link: https://github.com/egnascimento/productivefieldsdetection

Comments

thank you

Submitted by Muhammed Bilal C A on Thu, 01/18/2024 - 05:06