High-resolution RGB image dataset for wheat seed varietal identification and purity assessment

Area of Science:

• Agricultural Science
• Computer Vision
• Machine Learning

Background:

• Manual identification of wheat seed varieties is labor-intensive, time-consuming, and error-prone, hindering optimal yield.
• Existing computer vision and machine learning approaches require location-specific datasets, which are often limited or unavailable.
• Accurate seed purity and varietal identification are critical for sustainable agriculture and maximizing crop productivity.

Purpose of the Study:

• To address the lack of region-specific datasets for wheat seed varietal identification.
• To create and present a publicly available, high-resolution wheat seed image dataset for Pakistan.
• To facilitate the development of advanced AI models for accurate and efficient wheat seed analysis.

Main Methods:

• Collected high-resolution RGB images of three major Pakistani wheat varieties (Akbar-19, Dilkash-20, Urooj-22) under controlled lighting and angles.
• Ensured each variety comprised 125 pure seeds to create a robust and representative dataset.
• Collaborated with the Wheat Biotechnology Lab at the University of Agriculture, Faisalabad, Pakistan.

Main Results:

• A novel, publicly accessible dataset of Pakistani wheat seeds has been established.
• The dataset features high-resolution images crucial for training sophisticated machine learning models.
• The study highlights the importance of local varietal datasets for agricultural AI development.

Conclusions:

• The developed dataset is a valuable resource for researchers studying Pakistani wheat varieties and seed collection techniques.
• This initiative bridges the AI innovation gap in agriculture by providing region-specific data.
• The dataset promotes cooperation and enhances the reliability of research data for broader utilization in sustainable agriculture.