Global Wheat Detection 2020 Dataset Auto-Download

[glenn-jocher]

This PR adds the Global Wheat Detection 2020 dataset http://www.global-wheat.com/ to the YOLOv5 autodownload dataset list.

• COCO
• COCO128
• VOC
• Argoverse-HD
• VisDrone2019-DET
• GlobalWheat-2020
• KITTI
• SKU110 (grocery store items)
• CityScapes
• Objects365
• OpenImages

# Global Wheat 2020 dataset http://www.global-wheat.com/
# Train command: python train.py --data GlobalWheat2020.yaml
# Default dataset location is next to YOLOv5:
#   /parent_folder
#     /datasets/GlobalWheat2020
#     /yolov5

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

New dataset configuration added for Global Wheat 2020, and filename capitalization fixed for VisDrone dataset.

📊 Key Changes

• Global Wheat 2020 Dataset Integration: A new configuration file has been added for the Global Wheat 2020 dataset.
• VisDrone Dataset Filename Update: The visdrone.yaml file has been renamed to VisDrone.yaml to adhere to consistent naming conventions.
• Dataset Download Script Updates: Streamlined import statements in the dataset download sections for both datasets.

🎯 Purpose & Impact

• Enhanced Dataset Support: With the addition of Global Wheat 2020 dataset configuration, users can now train models specifically on this dataset, which is particularly useful for agricultural and plant surveillance applications.
• Improved Consistency: Correcting the filename capitalization of the VisDrone dataset ensures consistency in the dataset naming scheme, reducing possible confusion or errors when referencing datasets.
• Simplified Code: The changes in the download script are aimed to make the code cleaner and more maintainable, which benefits developers working on or extending the project.

[glenn-jocher]

labels.jpg

train_batch0.jpg