Datasets:
dataset_info:
features:
- name: image
dtype: image
- name: objects
struct:
- name: bbox
list:
list: float64
- name: categories
list:
class_label:
names:
'0': single_seedling
'1': double_seedling
'2': triple_seedling
- name: flight_date
dtype: string
- name: source_video
dtype: string
- name: split
dtype: string
splits:
- name: train
num_bytes: 15545420923
num_examples: 3152
download_size: 15545391709
dataset_size: 15545420923
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
license: unknown
task_categories:
- object-detection
size_categories:
- 1K<n<10K
Maize Seedling Msdd Detection
The Maize Seedling Detection Dataset (MSDD) is a stand-counting dataset of maize genetic nurseries at the University of Missouri in Columbia, Missouri, made from RGB video frames that DJI Phantom 4 and Mavic 2 Pro drones recorded over the 2019-2022 growing seasons. It covers seedlings from growth stage V2 to V12 under varied lighting, wind, soil backgrounds, and nadir or oblique camera angles. Each seedling is boxed as a single plant, or as a double or triple cluster when two or three plants emerge together. The flight_date and source_video columns record which flight and video each frame patch came from, and the split column keeps the authors' train/val/test split. The dataset contains 3,152 images with 163,921 bounding box annotations across 3 categories.
This dataset is indexed on https://project-agml.github.io/ as part of the AgML python library.
The original train/test/val split has been preserved in the split column.
Citation
@article{kharismawati2026maize,
title={Maize Seedling Detection Dataset (MSDD): A Curated High-Resolution RGB Dataset for Seedling Maize Detection and Benchmarking with YOLOv9, YOLO11, YOLOv12 and Faster-RCNN},
author={Kharismawati, Dewi Endah and Kazic, Toni},
journal={Agronomy},
volume={16},
pages={1605},
year={2026},
publisher={MDPI}
}
This dataset was reformatted from its original format to match HuggingFace standards.