WinterCropWeedDB-Sample: A Benchmark Dataset for Weed Classification in Winter Crops

This repository provides a sample version of the WinterCropWeedDB dataset, designed for weed classification research in winter crop fields. The complete dataset contains 1136 high-resolution images of six major winter crops (Wheat, Chickpea, Pea, Lentil, Mustard, Grass Pea) and four prevalent weed species (Common Vetch, Lesser Canary Grass, Goosefoot, Euphorbia Clementei).

In this sample release, we provide:
35 images per crop class
20 images per weed class

The images were collected directly from winter crop fields in Bilaspur and Mungeli districts, Chhattisgarh, India, using a 50 MP smartphone camera in Pro Mode under diverse lighting conditions (sunlight, shade) and camera perspectives to reflect real-world variability.

This subset provides a structured, annotated, and high-resolution resource that can be used for testing classification models, developing machine learning pipelines, or conducting preliminary research in weed detection, smart spraying, and sustainable crop management.

⚠️ Note: This is a sample release. The full dataset of 1136 images will be made available in future versions.

Data format: images.zip → Contains all images in .jpg format, organized by class folders
Use cases: Crop–weed classification, precision agriculture, machine learning benchmarking
Geographic Location: Bilaspur & Mungeli, Chhattisgarh, India

Value of the Data (Sample Version):

-Offers an initial benchmark for model testing and reproducibility

-Supports machine learning applications in smart agriculture

-Demonstrates variability in lighting, angle, and crop–weed morphology

本仓库提供了WinterCropWeedDB数据集的样本版本，该数据集专为冬田杂草分类研究设计。完整数据集包含1136张高分辨率图像，涵盖6种主要冬作物（小麦（Wheat）、鹰嘴豆（Chickpea）、豌豆（Pea）、兵豆（Lentil）、芥菜（Mustard）、草豌豆（Grass Pea））以及4种常见杂草（普通野豌豆（Common Vetch）、小雀麦（Lesser Canary Grass）、藜（Goosefoot）、克莱门特大戟（Euphorbia Clementei））。
在本次样本发布中，我们提供如下内容：每个作物类别35张图像，每个杂草类别20张图像。
这些图像采集自印度恰蒂斯加尔邦比拉斯布尔（Bilaspur）与蒙格利（Mungeli）地区的冬田，采用5000万像素智能手机的专业模式拍摄，涵盖多样光照条件（自然光、遮阴环境）与拍摄视角，以还原真实田间场景的多样性。
该子集为结构化、带标注的高分辨率资源，可用于分类模型测试、机器学习流水线开发，或开展杂草检测、智能喷施与可持续作物管理等领域的初步研究。
⚠️ 注意：本版本仅为样本发布。包含1136张图像的完整数据集将在后续版本中公开。
数据格式：images.zip → 包含所有以.jpg格式存储的图像，按类别文件夹进行组织。
应用场景：作物-杂草分类、精准农业、机器学习基准测试
地理位置：印度恰蒂斯加尔邦比拉斯布尔与蒙格利地区
样本数据集的价值：
- 为模型测试与研究可复现性提供初始基准
- 支撑智能农业领域的机器学习应用
- 展现光照条件、拍摄角度与作物-杂草形态的多样性