argus-cv 1.8.3

CLI tool for working with vision AI datasets

argus-cv

Vision AI dataset toolkit for working with YOLO, COCO, and semantic mask datasets.

Documentation

Library users can start with the Python API reference:

• https://pirnerjonas.github.io/argus/reference/python-api/

Installation

uvx argus-cv

Highlights

• Detects YOLO, COCO, and folder-based semantic mask datasets.
• Reports per-class stats (pixel coverage for mask datasets).
• Interactive viewer with bounding boxes, polygons, or mask overlays.
• Split unsplit YOLO/COCO datasets into train/val/test.

Usage

# List datasets in current directory
uvx argus-cv list

# List datasets in specific path
uvx argus-cv list --path /path/to/datasets

# Limit search depth
uvx argus-cv list --path . --max-depth 2

# Show instance statistics for a dataset
uvx argus-cv stats --dataset-path /path/to/dataset

# Short form
uvx argus-cv stats -d /path/to/dataset

# View annotations interactively
uvx argus-cv view -d /path/to/dataset --split val

# Split an unsplit dataset into train/val/test
uvx argus-cv split -d /path/to/dataset -o /path/to/output -r 0.8,0.1,0.1

# Filter dataset to selected classes
uvx argus-cv filter -d /path/to/dataset -o /path/to/output --classes person,car

# Convert mask -> YOLO segmentation
uvx argus-cv convert -i /path/to/masks -o /path/to/output --to yolo-seg