BanglaPulse: A Real-World Multiclass Image Dataset for Bangladeshi Pulse Crop Recognition
- Source
- DataCite
- Published
- 2026
- Country
- Bangladesh
- License
- Creative Commons Attribution 4.0 International
- Themes
- Agriculture & Farming, Machine Learning & AI, Remote Sensing & Satellites
About this dataset
Pulses are nutritionally vital and environmentally sustainable crops that play a central role in food security across South Asia, particularly in Bangladesh, where they serve as a primary source of plant-based protein for millions of people. Despite their agricultural and nutritional significance, publicly accessible benchmark image datasets representing Bangladeshi pulse crop varieties under real-world conditions remain scarce, limiting the development of AI-driven smart agricultural systems for this region. To address this gap, we present a real-world object detection dataset comprising 2,942 high-resolution images across 16 pulse crop categories, including Black Gram, Grass Pea, Pigeon Pea, Horse Gram, Chickpea (polished and unpolished), Red Lentils, Split and Husked Green Gram, Split Black Gram, Unpolished Grass Pea, Yellow Split Pea, White Pea, Green Gram, Hyacinth Bean, and...
How the data was collected
Images were collected from traditional markets in Dhaka, Bangladesh (Kawran Bazar and Dakshinkhan Kacha Bazar) between August and September 2025 using five different smartphone devices under varying illumination conditions, camera angles, shooting distances, and background environments.