mcedrdiego / tofu_video_labeller

Video Labeller tool

Tofu - SmartOps Video Labelling Tool

This is a minimalist tool for labelling video clips, e.g. activities, in the context of Deep Learning applications. So far, it exports the labelled slices to a csv file.

Installation

Ubuntu

If you are planning to play MP4 or any other proprietary formats, you need to install the following packages:

$ sudo apt install ubuntu-restricted-extras
$ sudo apt install build-essential qt5-default
$ sudo apt install libgstreamer1.0-0 gstreamer1.0-plugins-base gstreamer1.0-plugins-good gstreamer1.0-plugins-bad gstreamer1.0-plugins-ugly gstreamer1.0-libav gstreamer1.0-doc gstreamer1.0-tools gstreamer1.0-x gstreamer1.0-alsa gstreamer1.0-gl gstreamer1.0-gtk3 gstreamer1.0-qt5 gstreamer1.0-pulseaudio

Then, install the required Python Libraries. It is recommended to create a virtual environment for the application, instead of installing it globally:

$ cd <project_dir>
$ python3 -m venv venv
$ source venv/bin/activate
$ pip install -U pip
$ pip install -r requirements.txt

Windows

To play videos windows you must install video codecs first.

1. Download here: LAV Filters

2. From latest releases download tofu_video_labeller.zip [Here]

3. Unzip file and run "Tofu_Video_Labeller.exe"

Usage

In order to open the application, go to the project's root path and run the executable:

$ cd <project_dir>
$ ./tofu

Default shortcuts

• Space: Play/Pause
• Arrow Up/Down: speed up/slow down video
• Arrow Right/Left: jump 10 seconds forwads/backwards

Opening videos

Open a video on File menu.

Creating labels

Add labels and set shortcuts for each one of them.

Labelling

Press the shortcut when the video is running or paused to label a slice of the video. It's required to press the shortcut again to mark the end of the labelled video clip. Labels can overlap each other and those which do not have the end mark are highlighted on the bottom right table.

Exporting

After labelling the video, you can export your tags to a csv file.