⇄ Format converter

Convert COCO JSON to CVAT XML

Upload a COCO JSON annotation file — the tool converts it to CVAT XML ready for import into CVAT. Processing runs on the server; your file is not stored.

Bounding boxes (bbox) and polygons supported
Class labels and object attributes preserved
Multiple images in a single file
No file storage — real-time processing
Free, no registration required

Convert file

COCO JSON supported (instances_*.json)

📂
Drop your file here
or click to browse · JSON up to 50 MB
Annotation type
CVAT version
✓ Conversion completed successfully
⬇ Download CVAT XML

What are COCO and CVAT, and why convert?

📦

COCO JSON format

COCO (Common Objects in Context) is one of the most common annotation formats for computer vision. It is used in PyTorch, Detectron2, MMDetection, and other frameworks. The file contains images, annotations, and categories structures.

🏷️

CVAT XML format

CVAT (Computer Vision Annotation Tool) is a professional open-source annotation platform. Tasks are imported using XML that describes objects, attributes, labels, and image bindings.

🔄

When do you need conversion?

When a dataset is already labeled in COCO (e.g. exported from another tool) but needs review or further labeling in CVAT. Conversion lets you open existing annotations in CVAT and continue working.

Format comparison

📄 COCO JSON
{
  "images": [{
    "id": 1,
    "file_name": "img.jpg",
    "width": 1920,
    "height": 1080
  }],
  "annotations": [{
    "category_id": 1,
    "bbox": [100, 200, 300, 150]
  }]
}
📄 CVAT XML
<annotations>
 <image id="1"
   name="img.jpg"
   width="1920"
   height="1080">
  <box label="car"
   xtl="100" ytl="200"
   xbr="400" ybr="350"/>
 </image>
</annotations>

FAQ

Are my files stored on the server?
No. The file is uploaded, processed in server memory, and returned immediately. Nothing is saved or shared with third parties.
What is the maximum file size?
Up to 50 MB. For larger datasets, contact us — we can help with manual conversion or batch processing.
Are polygons (segmentation) supported?
Yes. Choose “Polygons” or “bbox + polygons” in the options. COCO segmentation (coordinate lists) is converted to CVAT <polygon> elements.
How do I import the result into CVAT?
In CVAT, open your task → Actions → Upload annotations → select “CVAT 1.1” or “CVAT 2.0” → upload the downloaded XML file.
Do I need to register?
No. The tool is completely free and does not require registration or sign-in.