Camera & Behavior¶
GLIDER can watch your experiment through a camera, record synchronized video and data, track animals or objects on screen, and score their behavior. This section walks through each of those capabilities in the order you would normally use them.
What's in this section¶
-
Connect a camera, watch the live preview, and record video alongside a set of frame-aligned CSV files. Also covers calibration (pixels to millimeters) and zones (regions of interest).
-
Detect and follow animals or objects with YOLO and ByteTrack, either live or as a batch pass over a recorded video. Requires the optional
visionextra. -
Label short video clips, train a classifier from your labels, and apply it to new videos to produce an ethogram. Requires the optional
behaviorextra.
Before you start¶
The camera and recording features work with a standard desktop install. Two of the more advanced capabilities live behind optional dependency groups, so you only install them if you need them:
| Capability | Extra | Install |
|---|---|---|
| Live preview + video/data recording | (included with pc) |
see Installation |
| YOLO / pose tracking | vision |
uv sync --extra vision |
| Behavior classification | behavior |
uv sync --extra behavior |
Video recording needs FFmpeg
Recording video requires FFmpeg to be installed on your system. If you followed the Installation guide you already have it. Preview, tracking, and data logging work without it.
Two different things are called 'Behavior Analysis'
GLIDER has a live, movement-based behavior classifier (built into the camera settings, described in Tracking) and a supervised machine-learning workflow (a separate window, described in Behavior Analysis). They are unrelated tools. The pages below point out which is which so you don't confuse them.