Tree Annotation for Satellite Images (YOLO Training Data)

I have 12,000 high-resolution satellite images that need precise tree annotations to train a YOLO-based deep learning model.
Each tree visible in the imagery must be outlined with an accurate bounding box and saved in YOLO Darknet text format (single class: tree).
You may use CVAT, QGIS, ArcGIS, or any other annotation tool you prefer — the key requirement is that the final files work flawlessly in a YOLO training pipeline.
=======================================================
Deliverables
• 12,000 annotation files (.txt), one per image, in YOLO Darknet format
• Each annotation file should include:
<class_id> <x_center> <y_center> <width> <height> (values normalized to image dimensions)
A short README file explaining the tools, settings, or export process used
=======================================================
Acceptance Criteria
• Every tree in every image must be annotated
• Missed or overlapping boxes should be under 5% in a random sample audit
• Annotation filenames must exactly match the source image names
• No formatting or syntax errors when tested in a YOLO training script
=======================================================
What to Include in Your Proposal
Please describe:
Your experience with large-scale image annotation projects, especially involving forestry or satellite imagery
The tools you plan to use (e.g., CVAT, LabelImg, ArcGIS, etc.)
Your expected throughput per day (number of images you can complete daily)
Any team members or methods you’ll use to ensure quality and consistency