Quick usage guide

Launching Pose-Trigger

  1. Open Terminal.

  2. Run the following command on Terminal:

    $ pose-trigger

Note

When being run without a parameter, Pose-Trigger will use the device at /dev/video0 by default. In case you want to use e.g. /dev/video1, specify the device as the parameter, i.e. run pose-trigger /dev/video1.

To check which path corresponds to your camera, please refer to the Appendix.

Organization of the main window

_images/Layout_overview.png

Overview of the main window

The Pose-Trigger main window can be divided into three groups:

  • The Capture buttons (yellow) are for starting/stopping acquisition.

  • The Preview panel (green) provides an on-line preview of the acquired video frames. If estimation of body-part positions is activated (refer to DeepLabCut evaluation), estimated positions will be shown as colored circles, too.

  • In the Settings panel (blue), you can configure how acquisition is performed (refer to the Panel-by-panel guide).

Capturing videos

Capture modes

There are two modes of running for Pose-Trigger:

  • FOCUS mode: capturing video frames without storing them

  • ACQUIRE mode: captures video frames and stores acquired data

You can start/stop either of the capturing modes by clicking on the button at the bottom of the main window.

Caution

Pose-Trigger does !not! stream data into storage during acquisition! During acquisition, it keeps all the data in-memory. The data will be written out to a file only after acquisition. The duration of acquisition will be thus limited to the order of 1–2 minutes.

Note

Currently, the following parameters are “hard-coded” and used as default:

  • Image format: 640x480 pixels, 16-bit grayscale

  • Timing generation: a busy-wait algorithm

  • Storage format: the NumPy zip-file format (.npz)

Format of the saved files

The data are saved in the NumPy zip-file format (i.e. “.npz” file). Each file includes the following entries:

Entries in saved files

Name

Always there?

Description

frames

Yes

the 3-D frame data, (frame-index, height, width)

timestamps

Yes

1-D array containing unix timestamps in seconds

metadata

Yes

a JSON-serialized text object containing information on acquisition configuration

estimation

No (Optional)

when a DeepLabCut project is selected; 3-D array with the (frame-index, parameter) shape

trigger_status

No (Optional)

when pose-evaluation is enabled; 1-D boolean array of evaluation results

TODO

add some examples for metadata (and probably for other entries, too)