Datumaro: Computer Vision Annotation and Management with Multi-Format Support

Open-edge-platform developed Datumaro, a Python-based framework for computer vision management, updated as of March 2026. It provides tools to build, analyze, and convert information across formats like COCO, YOLO, and Pascal VOC.

Use Cases

• Converting COCO annotations into YOLO format for object detection tasks
• Merging multiple Pascal VOC projects into a single unified structure
• Filtering ImageNet classification labels to extract specific object categories

Strengths

• Supports industry-standard formats including COCO, YOLO, and Pascal VOC
• MIT licensed for open use
• Includes both a Python API and a Command Line Interface

Limitations

• Contains no raw image files, acting only as a management layer for existing data
• Requires external data sources to utilize its analysis and conversion features

Provenance

Source

open-edge-platform

Collection Method

software development

Freshness

Updated as of March 2026.