DETECT, TRACK AND CLASSIFY OBJECTS IN YOUR CAMERA STREAMS OR FILES
Few solutions come close to the accurate, auto-detected contours in the Falcon Tracker application to identify, track and classify objects in video content.
AI and Deep learning-driven
Our software detects and classifies objects based on trained deep neural networks to enable smart video analysis, including search, filtering, data aggregation, visualization and alerting
Automatic Object Contour Detection
Application auto-detects accurate object contours to filter out noisy background objects like shadows, reflections, branches, waves, clouds and even nearby objects
Flexible Speed vs. Performance Tradeoffs
Application can switch between superior contour-based detection to traditional bounding-box detection by choosing different ML models at a differential compute cost
Coarse- and Fine- Grained Classification
Coarse-grained attributes such as object color, make/model of vehicles etc. are determined using appropriately trained ML models. Fine-grained pixel-level features are extracted using traditional signal processing for finer search and match
Tailored models to suit your needs
We have created and have the expertise to create and refine models for specific application use-cases. Single- or multi-label classification? Loss functions? Training data requirements? This is our bread-and-butter – talk to us and we can explore how to tailor models for your needs
Flexible deployment options (Edge and Cloud)
The Falcon Tracker application can run at the Edge (close to the camera) on ARM CPUs + NVIDIA Jetson GPUs and on-prem/in the Cloud on Intel/AMD CPUs + NVIDIA server GPUs.
Falcon Tracker
State-of-the-art detection, tracking, and fine-grained classification of objects.
AI-driven, containerized, plug-and-play with cameras
Demo: Key features of Falcon Tracker
See it in action: control cameras, view the camera stream on any device, view detailed statistics – ready to use!
Contours and Region-of-Interest
Fine-grained Classification
