# MSDD: Maize Seedling Dataset for Stand Counting

## Overview
The MSDD (Maize Seedling Dataset) is designed for plant detection and stand counting tasks in agricultural environments.  
It contains aerial imagery of maize genetic nurseries with annotations for early-stage seedling detection and classification.

The dataset is provided in YOLO format and supports training with modern object detection frameworks such as YOLOv9, YOLOv10, and YOLOv11/12.

---

## Directory Structure

```
yolo_format/
├── training/
│   ├── images/
│   └── labels/
├── validation/
│   ├── images/
│   └── labels/
├── test/
│   ├── images/
│   └── labels/
├── yolov9_data.yaml
├── yolov11-v12_data.yaml
└── README.md
```

### Description

- `training/images/` — training images  
- `training/labels/` — YOLO annotations for training images  

- `validation/images/` — validation images  
- `validation/labels/` — YOLO annotations for validation images  

- `test/images/` — test images  
- `test/labels/` — YOLO annotations for test images  

- `yolov9_data.yaml` — configuration file for YOLOv9 training  
- `yolov11-v12_data.yaml` — configuration file for YOLOv11/YOLOv12 training  

---

## Annotation Format

Annotations follow the standard YOLO format.

Each label file contains one object per line:

```
<class_id> <x_center> <y_center> <width> <height>
```

- Coordinates are normalized to the range `[0, 1]`
- Each `.txt` file corresponds to one image with the same filename

---

## Classes

| Class ID | Description |
|--------|------------|
| 0 | single |
| 1 | double |
| 2 | triple |

---

## Dataset Usage

### Example (YOLOv9)

```bash
yolo task=detect mode=train model=yolov9-c.pt data=yolov9_data.yaml
```

### Example (YOLOv11 / YOLOv12)

```bash
yolo task=detect mode=train model=yolo11.pt data=yolov11-v12_data.yaml
```

---

## YAML Configuration Files

### yolov9_data.yaml
```yaml
train: <path_to_dataset>/training/images
val: <path_to_dataset>/test/oblique/images
test: <path_to_dataset>//test/all/images

nc: 3
names: 
  0: Single
  1: Double
  2: Triple
```

### yolov11-v12_data.yaml
```yaml
path: <path_to_dataset>/

train: training/images
val: test/shadow/long/images
test: test/all/images

names: 
  0: Single
  1: Double
  2: Triple
```

---

## Citation


This dataset is associated with a manuscript under review and is currently available as a preprint.

If you use this dataset, please cite:

Dewi Endah Kharismawati and Toni Kazic.  
*Maize Seedling Detection Dataset (MSDD): A Curated High-Resolution RGB Dataset for Seedling Maize Detection and Benchmarking with YOLOv9, YOLO11, YOLOv12 and Faster-RCNN*.  
arXiv preprint arXiv:2509.15181, 2025.  
DOI: 10.48550/arXiv.2509.15181

### BibTeX

```bibtex
@article{kharismawati2025msdd,
  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={arXiv preprint arXiv:2509.15181},
  year={2025},
  doi={10.48550/arXiv.2509.15181}
}
```

---

## License

Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)

https://creativecommons.org/licenses/by-nc-sa/4.0/

---

## Contact

Dewi Endah Kharismawati, Ph.D
kharismawati.2@osu.edu