Individual trajectories per animal per frame
X/Y coordinates, velocity, and heading for each identified animal at every frame. Identity labels are consistent across the entire session.
CSVJSONMP4
← Capabilities
Multi-Animal Tracking
Markerless identification of individual animals in group-housed paradigms
Track up to four animals simultaneously with persistent identity assignment — no ear tags, paint marks, or RFID implants.
98.7%
Identity accuracy
30fps
Acquisition frame rate
4
Simultaneous subjects
24
Supported paradigms
The problem
Physical markers introduce confounds and limit throughput
Group-housed behavioral experiments require individual identification. Current methods — paint marks, ear tags, RFID implants — each introduce their own confounds. Paint degrades within sessions, ear tags cause grooming artifacts, and implants require surgery. Manual scoring of social interactions is subjective and prohibitively slow for large cohorts.
- Paint markers degrade during sessions, causing identity swaps and data loss in later time bins
- RFID implants require invasive surgery, adding stress and post-operative recovery confounds
- Manual scoring of social interactions is subjective, slow, and cannot capture all simultaneous events
The solution
Persistent identity from appearance alone
ConductVision assigns and maintains individual identities using appearance-based features extracted directly from video. Identity persists through occlusions, crossings, and huddling. All pairwise interactions are scored automatically and continuously.
- Persistent identity assignment without any physical markers — purely vision-based
- Identity maintained through occlusions, crossings, and close proximity at 98.7% accuracy
- All pairwise interactions scored automatically — approach, contact, following, avoidance
Endpoints
Per-animal dependent variables in group paradigms
Social proximity metrics
Inter-animal distances, contact durations, approach/withdrawal counts, and following behavior scored for every dyad continuously.
CSV
Zone occupancy by individual
Time in each defined zone reported per animal. Enables individual preference indices in social choice paradigms.
CSV
Applications
Group-housed paradigms requiring individual identification
Three-chamber sociability
Track the test animal and stimulus animals independently. Compute social preference index, investigation time, and approach frequency without manual scoring.
Measures
- Social preference index
- Investigation duration
- Chamber transition count
Resident-intruder aggression
Identify resident and intruder throughout the session. Score attack latency, bout duration, and inter-animal distance without observer bias.
Measures
- Attack latency
- Aggression bout duration
- Inter-animal distance
Observer-demonstrator learning
Track observer and demonstrator separately to quantify observation time, proximity during demonstration, and subsequent behavioral mimicry.
Measures
- Observation duration
- Proximity to demonstrator
- Behavioral mimicry index
Group open field phenotyping
Simultaneously phenotype multiple animals in the same arena. Individual locomotor profiles, thigmotaxis, and social clustering are computed in a single session.
Measures
- Individual distance traveled
- Thigmotaxis per animal
- Social clustering index
Compared to typical systems
How ConductVision differs
| Feature | ConductVision | Typical systems |
|---|---|---|
| Subjects tracked simultaneously | 4 — markerless | 2 with physical markers |
| Temporal resolution | 30 fps per animal | 8-11 fps |
| Identity through occlusion | Yes — 98.7% accuracy | Frequent identity swaps |
| Data granularity | Per-animal per-frame coordinates | Aggregate or manual annotation |
| Scoring method | Fully automated, all dyads | Manual or semi-automated |
Related capabilities
Social Interaction
Continuous automated scoring of all pairwise social interactions — approach, contact, following, avoidance — for up to 4 animals.
Unsupervised Discovery
Unsupervised behavioral clustering reveals structure without predefined ethograms — UMAP visualization and novel behavior flagging.
Gait Analysis
HMM-based stride detection from standard video with per-paw metrics — no treadmill or pressure plate required.
Try it with your own recordings
Upload a multi-animal recording and see individual trajectories extracted without markers.
