greenhouse_crop_weed_detection
收藏资源简介:
该数据集名为 Greenhouse Crop Weed Detection,是一个用于温室场景下作物与杂草检测的目标检测数据集。数据集包含200张图像,共标注了11,192个边界框,覆盖14个类别,包括:Blackbean、Canola、Corn、Field Pea、Flax、Horseweed、Kochia、Lentil、Palmer Amaranth、Ragweed、Redroot Pigweed、Soybean、Sugar beet 和 Waterhemp。数据集的标注格式为边界框(bbox)和类别标签(categories)。该数据集源自 AgML 项目,原始数据经过重新格式化以符合 HuggingFace 标准。许可协议为 CC-BY-4.0。
The dataset is called Greenhouse Crop Weed Detection, an object detection dataset for crop and weed detection in greenhouse scenarios. It contains 200 images with 11,192 bounding boxes covering 14 categories: Blackbean, Canola, Corn, Field Pea, Flax, Horseweed, Kochia, Lentil, Palmer Amaranth, Ragweed, Redroot Pigweed, Soybean, Sugar beet, and Waterhemp. The annotation format is bounding box (bbox) and category labels. The dataset originates from the AgML project, and the original data has been reformatted to comply with HuggingFace standards. The license is CC-BY-4.0.
数据集概述
Greenhouse Crop Weed Detection 是一个温室环境下农作物与杂草检测数据集,支持目标检测任务。
基本信息
- 许可证: CC-BY-4.0
- 任务类型: 目标检测(Object Detection)
- 数据集规模: 少于 1,000 个样本(n<1K)
- 数据来源: 原始格式已调整为 HuggingFace 标准格式;数据集收录于 Project-AgML 并作为 AgML Python 库的一部分
数据内容
- 图像数量: 200 张
- 边界框注释数量: 11,192 个
- 类别数量: 14 类,包括:
- 农作物(7 类):Blackbean(黑豆)、Canola(油菜)、Corn(玉米)、Field Pea(田豆)、Flax(亚麻)、Lentil(扁豆)、Soybean(大豆)、Sugar beet(甜菜)
- 杂草(6 类):Horseweed(加拿大蓬)、Kochia(地肤)、Palmer Amaranth(帕尔默苋)、Ragweed(豚草)、Redroot Pigweed(红根苋)、Waterhemp(水麻)
数据划分
- 训练集(train): 200 个样本,大小约 2.80 GB(2,802,533,096 字节)
- 下载大小: 约 2.80 GB(2,802,554,800 字节)
数据特征
- image: 图像数据
- objects: 目标对象信息,包含:
- bbox: 边界框坐标(float64 列表)
- categories: 类别标签(14 个类别,从 0 到 13)
引用信息
引用论文:
Sunil, GC, Koparan, Cengiz, Upadhyay, Arjun, Ahmed, Mohammed Raju, Zhang, Yu, Howatt, Kirk, Sun, Xin (2024). "A novel automated cloud-based image datasets for high throughput phenotyping in weed classification." Data in Brief, 57, 111097.
引用数据集:
G C, Sunil; Koparan, Cengiz; Upadhyay, Arjun; Ahmed, Mohammed Raju; Zhang, Yu; Howatt, Kirk; Sun, Xin (2024), "A Novel Automated Cloud-Based Image Datasets for High Throughput Phenotyping in Weed Identification." Mendeley Data, V3, doi: 10.17632/hs7d7kpd3z.3




