Features
Our technology combines advanced AI models with flexible deployment to deliver reliable, real-time object tracking across cameras. Below are the core components that make our solution accurate, scalable, and ready for real-world use:
Visual Input - Natural image sequences from any source, including video or multi-view datasets
Model-Agnostic Engine - Compatible with pinhole, fisheye, and custom lens models across varying resolutions and sensor types.
Robust Estimation - Self-supervised optimization pipeline handles noise, motion blur, and unstructured environments
Scalable Deployment - Lightweight and automation-friendly, ideal for large-scale processing and integration into existing data pipelines
Target-Free Calibration - No need for checkerboards or calibration rigs—works directly from natural image sequences.
See how it works
Here is an example of the intrinsic calibration of an infrared (IR) camera.
visionairy®
Auto Intrinsic Camera Calibration
Automatic intrinsic camera calibration estimates a camera's internal parameters—like focal length and lens distortion—without requiring calibration targets, enabling fast and flexible setup across diverse optics.
These performance metrics are for demonstrative purposes only, based on configurations with proven results. Actual performance may vary by setup. Our algorithms are optimized for use with any chip, platform, or sensor. Contact us for details.
Reprojection Error
ca. 1 px in typical environments
Processing time
ca. 15 min (depends on calibration sequence)
Supported companion hardware
Supported companion hardware: Nvidia Jetson Orin AGX, AMD64 with Nvidia graphic
Basis-SW/OS
Linux, Docker required
Interfaces
ROS2
Input - Sensors
• Any type of camera (sensor agnostic)
• Any type of GPS
Input - Data
1) Camera's video frames
2) GPS measurements
Output - Data
Intrinsics for the following camera models:
• Double Sphere
• Radtan
• EquidistantMinimum
Recommended
RAM
16 GB
32 GB
Storage
20 GB
50 GB
Camera
640 x 480 px, 15 FPS
1920 x 1080 px, 30 FPS
GPS
-
X
The information provided reflects recommended hardware specifications based on insights gained from successful customer projects and integrations. These recommendations are not limitations, and actual requirements may vary depending on the specific configuration.
Our algorithms are compatible with any chip, platform, sensor, and individual configuration. Please contact us for further information.
Performance Metrics
Position accuracy
±2.5 cm in typical environments
update rate
Up to 200 Hz
initialization time
<1 second
Maximum Velocity
20 m/s with full accuracy
Operating Range
Unlimited (environment-dependent)
Drift
<0.1% of distance traveled
Benefits
Target-Free
Calibrate from real-world footage, no checkerboards or special setups required.
Camera-Agnostic
Works with any camera model, lens type, or resolution, including fisheye and wide-angle.
Ready for the Wild
Perfect for in-the-wild data, archival footage, or systems where traditional calibration is not an option
Why Auto Intrinsic Camera Calibration?
Traditional intrinsic calibration methods rely on checkerboards and controlled setups, which are time-consuming and often unworkable for large-scale, mobile, or legacy image collections. Their rigid requirements limit use in dynamic or real-world environments.
Automatic intrinsic calibration removes these barriers by estimating camera parameters directly from natural image sequences. It works seamlessly with existing data, enabling scalable, fast calibration across varied platforms and conditions.
Enables Real-Time and On-the-Fly Calibration
Reduces Operational Overhead
Scales easily across diverse camera types and conditions
