Lightning-fast behavioral scoring tool.
Scoring behavior is time-consuming. Verifying pre-identified behavior is much faster.
This application takes 1) video, 2) event timestamps, and 3) a timeseries. The video is then clipped into 1-s segments for faster scoring.
This application took my behavioral scoring from several hours to ~20 minutes per subject.
Hotkeys are paramount here. Go slow until they are second nature.
-
Clone the repo
-
(Recommended) Create a virtual environment
python -m venv .venv -
Install dependencies
pip install -r requirements.txt
(If you installed the dependencies in a virtual environment):
source .venv/bin/activate # macOS
. .\.venv\Scripts\activate # Windows
Launch the app:
`python startup.py`
Open the files:
- Click
1to open a video file - Click
2to open an events file (single-column CSV, no header row) - Click
3to open a timeseries file (single-column CSV, no header row)
Simply move the center line in the timeseries graph to adjust the timing of the current event.
If you see an event that is not in your original CSV, but you want to add it to the data, hold down Alt and move
the center line on the TimeSeries graph to the desired position.
Click Shift to open the label editor. Use tab to toggle left/right, and type the label. The autosuggestions can
be changed here. Click Enter to save the label.
Saving is performed automatically with every operation.
# Files
1: Open Video
2: Open Events
3: Open Timeseries
# Playback
Left Arrow: Previous Event
Right Arrow: Next Event
Right Mouse Click: Next Event
Space: Replay current event
Alt: Hold and drag the center line to create a new event
# Playback Speed
Comma: 1x Speed
Period: 0.5x Speed
Slash: 0.25x Speed
# Events
F: Flag event
D: Discard event
# Labels
Shift: Open label editor
Tab: Toggle L/R prefix
Enter: Save label and close editor
QuickScore is distributed under the MIT License.
