carloscaetano/YoloToKitti

A converter from Yolo to Kitti detections

Yolo to Kitti Converter (YoloToKitti)

This repository holds the YoloToKitti. A simple code to convert Yolo detection files to Kitti files.

• Yolo format: (int_label, x_center, y_center, width, height) - float values relative to width and height of image, it can be equal from (0.0 to 1.0].
• Kitti format: (str_label, x1, y1, x2, y2) - "real" bounding box values.

Contents

• Usage Guide
  • Prerequisites
  • Code & Data Preparation
    • Get the code
    • Usage
• Other Info
  • Contact

---

Usage Guide

Prerequisites

[back to top]

The main dependency to run the code is

• OpenCV

The codebase is written in Python 3.6. We recommend the Anaconda Python distribution.

Code & Data Preparation

Get the code

[back to top]

Use git to clone this repository

git clone --recursive https://github.com/carloscaetano/YoloToKitti

Usage

[back to top]

To convert the Yolo format detection files to Kitti format, run the YolloToKitti.py. It has two arguments:

• [--detections, -d] Directory containing the Yolo detection files (.txt)
• [--images, -i] Directory containing the corresponding image files (.jpg)
• [--output, -o] Directory to save the converted Kitti detection files (.txt)

Example

To convert the Yolo detection files, at the example directory ./yoloV4_detections, with the corresponding image files at ./images and save the converted Kitti detection files to the folder ./files_converted, you can run

python YolloToKitti.py --detections ./yoloV4_detections --images ./images --output ./files_converted

or

python YolloToKitti.py -d ./yoloV4_detections -i ./images -o ./files_converted

Other Info

[back to top)]

Contact

For any question, please contact

Carlos Caetano: carlos.caetano@dcc.ufmg.br