Turning noisy sensor data into actionable tracks.
Estimum Technology is a 100% Canadian company specializing in AI-based algorithms for estimation theory, target tracking, video tracking and sensor fusion — powering modern surveillance and smart-city systems from raw detections to confident, continuous situational awareness.
AI-based algorithms across four disciplines
We specialize in developing artificial intelligence and machine learning algorithms for the full signal-to-decision chain — deep networks where they earn their keep, principled estimation where they do not, and both designed together rather than bolted together.
Estimation Theory
Learned motion and measurement models fused with Kalman, unscented and particle filters — neural state estimation that stays grounded in rigorous statistics.
Target Tracking
Multitarget, multisensor tracking with AI-driven data association and learned clutter rejection — reliable track initiation and maintenance in dense scenes.
Video Tracking
Deep-learning detection and re-identification across occlusion, with real-time neural inference optimized for embedded and edge hardware.
Sensor Fusion
Radar, EO/IR, lidar and video fused by learned models at measurement or track level, delivering one consistent, calibrated and uncertainty-aware world view.
Where our algorithms go to work
The same estimation core adapts to very different operating environments.
Smart Cities
Understanding how people and vehicles move through urban space — accurately, continuously, and at city scale.
- Traffic flow and intersection analytics
- Pedestrian and cyclist safety monitoring
- Multi-camera tracking across coverage gaps
- Infrastructure planning from movement data
Surveillance
Persistent wide-area awareness where targets are sparse, sensors disagree, and clutter is the norm.
- Wide-area and perimeter surveillance
- Maritime, ground and air target tracking
- Counter-UAS detection and track fusion
- Threat assessment and track classification
AI that answers to the mathematics
Learning where it wins, theory where it counts
Neural networks are exceptional at perception — detecting, describing and re-identifying objects. Estimation theory is exceptional at reasoning over time under uncertainty. We develop AI-based algorithms that combine the two, so a deep model never gets the last word on a decision it cannot justify.
Explainable, and shipped
Our models report calibrated uncertainty rather than bare confidence scores, so when a track breaks we can say exactly why. Prototypes become tested, instrumented, real-time software, and we stay alongside your team through integration, field trials and tuning.
Have a tracking problem worth solving?
Tell us about your sensors, your targets and your environment. We will tell you honestly what is achievable.
info@estimum.com