Merged event timeline
Single file containing automated tracking results and manual annotations with identical timestamps — no post-hoc alignment.
CSVJSON
← Capabilities
Annotation and Manual Scoring
Manual annotation tools integrated with automated tracking
Mark custom behavioral events with configurable keyboard shortcuts directly on the tracking timeline — automated and manual scores merged in a single time-aligned output.
33ms
Annotation precision (1 frame)
Unlimited
Custom event types
Yes
Auto + manual merge
Keyboard
Configurable shortcut scoring
The problem
Separate annotation tools break the analysis workflow
When automated tracking cannot score a behavior, researchers switch to separate annotation tools (BORIS, Solomon Coder, Noldus Observer) that produce disconnected event logs. Merging manual annotations with automated tracking requires custom alignment scripts.
- Switching between tracking software and annotation tools wastes time and loses context
- Manual annotations and automated tracking use different time bases — alignment is error-prone
- Separate tool outputs must be merged post-hoc with custom scripts
The solution
Integrated annotation on the tracking timeline
ConductVision includes a built-in annotation panel with configurable keyboard shortcuts. Press a key to mark event onset, release for offset. Manual events appear on the same timeline as automated tracking data — no alignment needed.
- Configurable keyboard shortcuts: assign any key to any behavioral event type
- Press-to-start, release-to-end marking for state events; single press for point events
- Merged output: automated tracking data and manual annotations in a single time-aligned file
Endpoints
Annotation outputs
Annotation-only export
Manual annotations exported separately with event type, onset, offset, duration, and frame numbers.
CSVJSON
Inter-rater reliability
When multiple users annotate the same recording, compute Cohen's kappa and percent agreement for scoring validation.
CSV
Applications
Annotation use cases
Cataloging uncommon events
Some behaviors (mounting, aggressive postures, seizure subtypes) are too rare for automated classifier training. Manual annotation captures them with frame-level precision.
Measures
- Event count
- Event duration
- Temporal distribution
Generating classifier training sets
Manual annotations serve as labeled training data for custom behavior classifiers — annotate examples, train the model, automate future scoring.
Measures
- Labeled frame count
- Class distribution
- Annotation consistency
Automated scoring validation
Manually score a subset of recordings to validate automated classifier accuracy — compute agreement between human and machine.
Measures
- Human-machine agreement
- False positive rate
- False negative rate
Multi-behavior social scoring
Social behaviors (mounting, chasing, grooming another) require contextual judgment. Manual scoring with keyboard shortcuts enables rapid multi-behavior annotation.
Measures
- Behavior frequency
- Bout duration
- Social event sequence
Compared to typical systems
How ConductVision differs
| Feature | ConductVision | Typical systems |
|---|---|---|
| Integration with tracking | Same timeline, no alignment needed | Separate tool, manual alignment |
| Keyboard shortcuts | Fully configurable | Fixed or limited shortcuts (BORIS) |
| Merged output | Auto + manual in single file | Separate outputs, custom merge |
| Inter-rater reliability | Built-in kappa calculation | External calculation |
| Frame-level precision | 33 ms (video frame rate) | Varies by tool |
Related capabilities
Ethogram Builder
Custom ethogram editor with hierarchical categories, exclusivity rules, and keyboard shortcut mapping.
Behavior Classification
ML-powered frame-level behavioral classification with temporal smoothing and custom classifier training.
Custom Model Training
Train and validate custom behavior classifiers on lab-specific annotated data.
Score what automation cannot — integrated in one tool
Set up keyboard shortcuts for your behavioral events and start annotating alongside automated tracking.
