360-object-detection-annotation

360ObjectAnnotator: a Tool to Annotate Object Bounding Box for 360 Videos

This is a tool to annotate bounding boxes for object detection in 360 videos. The tool provides a GUI to locate objects in equirectangular images via mouse interaction.

Requirements

This tool has been tested in Windows OS. It may not work on Linux. Python2 is required.

Installation

1) Install Anaconda 2) Create conda environment for python 2 3) Check requirements.txt for all required packages.

Usage

Create an images folder inside the root folder if it has not been created before. Remove all non images files such as .txt from the images folder.

Go into the root folder of this project and start the program using this command:

python main.py

After the program runs, the main window will appear showing content of the first image inside the images folder in equirectangular format. Move the mouse cursor to the center of the object to be annotated and left click. Another small window will pop up showing the viewport projection containing the object of interest.

Drag the mouse from top-right to bottom-left to create a bouding box. Type the designated key to assign the object class to the bounding box. Refer to config.ini file for key-class bindings.

Type a or d to move between images stored inside the ‘images’ folders Type z to undo the most recent action Type q to quit the program

Contact

If you have any questions, please use the public issues section on this GitHub. Alternatively, drop us an e-mail at jliao2@gmu.edu or anguy59@gmu.edu.

Ackowledgement

This tool is created with contributions from David Joy. A part of the code is reused from the Pano2Vid project.