Vegetable Object Detection Dataset from Bangladesh

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.