Browse categories
Explore
Fiverr Pro
English
$
USD
I will process and clean coco format annotation datasets for cvat
Muazam Ali
About this gig
Are you struggling with messy COCO-format annotation datasets that aren't ready for CVAT or model training? I can help.
I specialize in processing large-scale COCO-format polygon segmentation datasets cleaning, adjusting, and packaging them so they're 100% ready for direct import into CVAT or your training pipeline.
What I do:
Inward polygon buffering (precision adjustments using Python + Shapely)
Automatic bbox and area recalculation after every edit
Zero data loss every original annotation is preserved as a fallback, even in edge cases
Before/after QC visualizations so you can verify accuracy
Final packaging as clean, CVAT-compatible ZIP files
Why work with me:
- Eperience handling batches from hundreds to tens of thousands of annotations
- Reliable, deadline-driven delivery
- Clear communication throughout the project
- Competitive per-polygon rates with bulk discounts
Whether you have a small dataset or a massive annotation backlog, I'll make sure it's clean, accurate, and ready to use.
Message me before ordering if you have a custom dataset size or specific requirements happy to discuss!
#DataAnnotation #COCODataset #ComputerVision #CVAT #ImageAnnotation #MachineLearning #DataLabel
Get to know Muazam Ali
Muazam Ali
Full Stack Developer, Python Automation and Data Annotation Specialist
- FromPakistan
- Member sinceJul 2026
Languages
English
I'm a versatile freelancer skilled in three areas: building web interfaces with Next.js and React, developing backend apps with Python and Django, and running data annotation/QA pipelines for computer vision datasets. I use Selenium and Playwright for automation, and I've processed thousands of COCO-format annotations using Python and Shapely — including polygon adjustments, QC visualization, and packaging datasets for tools like CVAT. I deliver clean, well-documented, deadline-driven work. Let's discuss your project
FAQ
Q1: What annotation format do you work with?
I specialize in COCO-format JSON annotations, particularly polygon segmentation for categories like roofs, chimneys, skylights, and similar objects. If you use a different format, message me before ordering.
Q2: Will any of my annotations be lost or deleted?
No. My process never drops annotations. If a polygon adjustment causes an issue, the original geometry is always preserved as a fallback.
Q3: Can you handle very large datasets (10,000+ annotations)?
Yes, I've processed batches ranging from a few hundred to tens of thousands of annotations. For very large datasets, please message me first so I can confirm delivery time and pricing.
Q4: Will the output be ready to import directly into CVAT?
Yes, all deliverables are packaged as CVAT-compatible ZIP files, ready for direct import.
Q5: Do you provide proof that the processing worked correctly?
Yes, I include before/after QC visualizations so you can clearly verify the changes made to your dataset.
