Vegetable Object Detection Dataset from Bangladesh

Published: 6 June 2025| Version 1 | DOI: 10.17632/gnc4s3z2mf.1
Contributors:
Sabrina Jahan, Mohammad Rifat Ahmmad Rashid, B M Shahria Alam, Ishraque Manzur, Tawhidur Rahman, Raiyan Gani, Karib Shams, Md Miskat Hossain, Mahamudul Hasan

Description

Context & Motivation: * Vegetables play a crucial role in Bangladesh’s nutrition, economy, and food security. * Accurate vegetable identification supports efficient cultivation, inventory management, and smart agriculture. Dataset Composition: * Total of 3,534 high-resolution images. * Captured under real-world, natural conditions from roadside vendor stalls. * Images acquired using a Redmi Note 12 smartphone mounted on local vendor vehicles. Vegetable Classes: * 22 distinct classes covering diverse appearances and shapes. * Includes common Bangladeshi vegetables across varied lighting and background conditions. Annotation & Format: * Annotated via the Roboflow platform for precise bounding-box labels. * Provided in Pascal VOC format to facilitate object detection benchmarks.

Files

Institutions

  • East West University

Categories

Computer Vision, Object Detection

Licence