Multi-Class Segmentation Annotation Task

I have a collection of images that require pixel-accurate object outlines. Every person, vehicle and visible product in each frame must be traced with a closed polygon so the resulting dataset is ready for training instance-segmentation models.

Scope
• Objects to outline: people, all kinds of vehicles, and any consumer products that appear in the scene.
• Marking method: exact shape (polygon) around every instance—not bounding boxes and not whole-image tags.
• Tools: feel free to use CVAT, LabelMe, Labelbox, VGG Image Annotator or another platform that can export COCO JSON, Pascal VOC XML or PNG masks; the format can be confirmed together before you begin.

Deliverables
1. Fully annotated image set with polygons for every required class.
2. Exported annotation file(s) in the agreed format.
3. A brief read-me describing the class names, tool used and any assumptions made while labeling.

I will supply clear labeling guidelines, class definitions and a small test batch so we can align on quality before you proceed with the full set. Accuracy and consistency are more important than speed; annotations that miss objects or include stray points will be returned for correction.

Let me know which tool you plan to use and how you ensure pixel-level precision—then we can get started right away.