I will do 3d lidar point cloud annotation and cuboid labeling
Contributor AI ML LLM NLP
About this Gig
Need 3D cuboid labels for your LiDAR dataset? I provide manual point-cloud annotation following your object classes and labeling guidelines.
SERVICE
1) 3D cuboids around clearly identifiable objects
2) Class labels and agreed object attributes
3) Quality checks against your instructions
4) Delivery in one agreed export format
PACKAGE SCOPE
One item means one object cuboid in one point-cloud frame. The same object labeled in another frame counts again.
Basic: 100 cuboids across up to 20 frames
Standard: 300 cuboids across up to 60 frames
Premium: 750 cuboids across up to 150 frames
Both cuboid and frame limits apply. Packages cover independent-frame annotation. Persistent tracking IDs, sensor fusion, segmentation, and sparse or difficult objects require sample review and a custom quote.
Tool compatibility and the output schema must be confirmed before ordering.
Revisions cover adjustments within the agreed scope. New classes, changed guidelines, or additional data require a separate quote.
Please send sample frames, your class list, guidelines, preferred tool, output schema, and deadline before ordering.
Technique:
Manual
Tagging type:
Image
•
Video
My Portfolio
FAQ
What counts as one item tagged?
One item is one 3D cuboid around one object in one point-cloud frame. If the same object appears in 10 annotated frames, it counts as 10 items. Both the package’s cuboid and frame limits apply.
Is tracking included?
Listed packages cover independent-frame cuboid annotation. Consistent tracking IDs across frames, interpolation, and trajectory labeling require a custom quote after sample review.
Which tools and export formats do you support?
Please share your preferred platform, sample input files, and required output schema before ordering. I will confirm compatibility before accepting the project. Export support depends on the tool and task.
What determines the price?
The workload depends on cuboid count, frame count, object visibility, point density, attributes, and review requirements. Sparse objects, crowded scenes, sensor fusion, and additional tasks require sample review and a custom quote.
How do you check annotation quality?
Each package includes a review of class labels, cuboid placement, dimensions, orientation, and missed objects against the agreed guidelines. Please provide rules for occluded, truncated, or ambiguous objects.
What do revisions cover?
Revisions cover adjustments within the original scope and instructions. Errors against the agreed guidelines will be corrected. New classes, changed rules, or additional frames require a separate quote.
Can you handle larger datasets?
Send representative frames, the estimated object and frame counts, and your deadline. I will review the workload and confirm capacity, pricing, and delivery before accepting the order.
Can you sign an NDA?
Please share the NDA and data-handling requirements before ordering so I can review and confirm whether I can meet them. Your data will not be used in portfolio samples without permission.

