With Detekt, you can use Artificial Intelligence to automatically identify road signs, road defects, manhole covers, road markings, and more.
Detekt is compatible with georeferenced images, point clouds, and GIS data, offering extensive location-based features for diverse applications.
Detekt provides immediate access to millions of current images and LiDAR point cloud data worldwide.
The pipeline utilizes a three-stage instance segmentation process to handle mobile mapping data effectively.
The detection results - points and polygons - are converted from image coordinates into world coordinates using 3D lidar data or depth maps from stereoscopic images and GNSS data. This transformation provides world-space projected points and polygons, accurately pinpointing the location of detected objects or surfaces.
Objects detected within multiple images and times are fused into a single, unified detection. This aggregation enhances the robustness and accuracy of the detection process through the principle of multiple 'votes' confirming the presence of the same object.
Detections and location data are individually integrated into your internal processes & workflows.
Interact with and explore your detection results through our intuitive Detekt Viewer.
Export results as a shapefile or GeoJSON for analysis using GIS tools such as ArcGIS or QGIS.
We can process any image data, including videos, from various sources like planar and stereoscopic images, and 360° captures. Our expertise spans major brands like Teledyne LadyBug, Leica, Trimble, and Riegl. Calibration of any camera setup is crucial to enhance detection accuracy.
We provide millions of current images and LiDAR point cloud data globally, in partnership with TomTom. For further details, please contact our Sales team.
For accurate location determination and proper display in the Viewer, GNSS coordinates (such as GPS, GLONASS, Galileo, or BEIDOU) and image orientation are essential.
For evaluating GIS accuracy, we split the road surface into tiles of 0.25m x0.25m. A tile is considered correctly classified if the overlap of all the annotation classes and the detected classes within it exceeds a certain threshold.
Detekt utilizes various methods to precisely locate and determine the size of detected objects or surfaces from provided image depth data:
Detekt employs a proprietary method, known as object fusion, to integrate separate detections from various images and data sources. Our custom algorithm not only merges classification and location data but also applies logical weighting to enhance accuracy. Additionally, it tracks and compares identical objects over different times to monitor condition changes.
Our viewer incorporates a high-quality, scalable image anonymization tool that blurs faces, bodies, and license plates in panoramas, planar images, and videos.
Detekt is a cloud-based solution. We also offer a Docker container that can be deployed on your local machine or infrastructure.
Cloud deployments benefit from more frequent updates and improvements to our models and software.
To comply with local regulations, Detekt can be deployed in a local AWS data center. Explore all available AWS data centers.
Conformal Prediction is a statistical method that provides valid measures of uncertainty for model predictions by generating prediction intervals or sets that contain the true outcome with a specified probability, regardless of the underlying data distribution.
In Detekt, Conformal Prediction enhances the reliability of identifying road cracks, traffic signs, and other features by assigning confidence levels to each detection, helping users assess the trustworthiness of the results and make better-informed decisions.
By integrating Conformal Prediction, Detekt increases the trustworthiness of its analytics software, providing users with quantifiable uncertainty estimates for each prediction, which leads to improved decision-making and confidence in the software’s performance across various conditions.
.svg)
By clicking “Accept All”, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage and assist in our marketing efforts.
