Skip to content

This library allows reading and converting bounding box annotations in many popular formats

License

Notifications You must be signed in to change notification settings

ODAncona/bboxconverter

Repository files navigation

bbox logo

Python versions Total downloads Monthly downloads Python versions
Documentation Status

bboxconverter

bboxconverter is a Python library that enables seamless conversion of bounding box formats between various types and file formats. It provides an easy-to-use syntax for reading and exporting bounding box files.

Introduction

What is a bounding box?

Bounding boxes are a crucial component of object detection algorithms, which are used to identify and classify objects within an image or video. A bounding box is a rectangle that surrounds an object of interest in the image, and is typically represented by a set of coordinates that define the box's position and size.

Bounding box example

Various types and format

When you work with bounding box you have severals things to consider.

The bounding box could be stored in different types like:

  • Top-Left Bottom-Right (TLBR), (x_min, y_min, x_max, y_max)
  • Top-Left Width Height (TLWH), (x_min, y_min, width, height)
  • Center Width Height (CWH), (x_center, y_center, width, height)

Which are popular among different formats like :

  • COCO (Common Objects in Context)
  • Pascal VOC (Visual Object Classes)
  • YOLO (You Only Look Once)

Furthermore, the bounding box could be stored in different file formats like:

  • csv
  • xml
  • json
  • manifest
  • parquet
  • pickle

Installation

pip install bboxconverter

or

git clone https://github.com/ODAncona/bboxconverter.git
cd bboxconverter
poetry install

See the installation guide for more informations.

Usage

The goal of this library is to seamlessly convert bounding box format using easy syntax.

It should be a breeze like...

import bboxconverter as bc

# Input file path
input_path = './examples/example.csv'

# Output file path
output_path = './examples/output/example.json'

# Mapping between the input file and the bboxconverter format
bbox_map = dict(
    class_name='class',
    file_path='name',
    x_min='top_left_x',
    y_min='top_left_y',
    width='w',
    height='h',
    image_width='img_size_x',
    image_height='img_size_y',
)

# Read the input file
parser = bc.read_csv(input_path, mapping=bbox_map)

# Export the file to the desired format
parser.export(output_path=output_path, format='coco')
parser.export(output_path=output_path, format='voc')
parser.export(output_path=output_path, format='yolo')

Documentation

You can find the documention online at bboxconvert.readthedoc.io

Changelog

See the CHANGELOG file for details.

Contributing

Contributions are welcome! Please read the contributing guidelines first.

License

This project is licensed under the GPLV3 License - see the LICENSE file for details.

Acknowledgments