Automotive
Accelerate your digital roadmap
Our end-to-end AI Data Solutions enable the creation of high-quality datasets for the development of mapping, perception and prediction models at scale. With our computer vision experience and expertise, we can help you streamline your data pipeline for higher throughput.
Partner to some of the world’s leading automotive innovators
Object detection and localization
Object detection and localization
For multi-object detection and localization models, our platform offers annotation support for standard classes such as pedestrians, cyclists, cars, traffic signs, etc. We also support additional classes and attribute customization for complex mobility applications.
Object detection and tracking
Object detection and localization
Autonomous vehicle manufacturers collect data that spans several hours and different conditions to train advanced mapping and perception systems. Our platform's automated object interpolation and tracking capabilities significantly simplify video annotations.
Panoptic segmentation
Panoptic segmentation provides granular, pixel-critical information for advanced ML algorithms combining instance and non-instance annotations within a single frame. Further, unique instance IDs for each object of interest are assigned automatically to each segment.
3D point cloud segmentation
Leverage accurate point-wise segmentation of 3D point clouds to build high-performing self-driving models. Our platform is compatible with a wide range of lidars, including solid-state or flash lidars that produce dense points with richer data.
Sensor fusion via 2D-3D linking
Detect and track objects of interest across 2D images and 3D point clouds captured using complex sensor set up found in autonomous vehicles. Our platform has an auto-linking feature that accurately links objects in both 2D and 3D scenes across multiple frames.
Object movement tracking
Accurately determine object shapes and track object movements via a sequence of pixel-accurate points using a landmark or key points annotation tool. These datasets are especially useful to train autonomous driver assistant/monitoring systems.
Irregular shape detection
Our platform’s auto-pixel detection for popular classes used in autonomous driving datasets (e.g. pedestrians, car shapes, road or lane markings, signboards, etc.) helps to simplify the annotation of different shapes and coarse objects in images and videos via polygons.
Lane detection
Autonomous vehicles are trained in numerous scenarios to accurately detect drivable areas by detecting lane lines during the daytime, in cities or highways and high-traffic areas. Our platform provides a polyline tool to support lane detection in 2D and 3D scenes.
Image Annotation
Build high-quality image datasets using data annotation tools that support 2D/3D bounding boxes, polygons, polylines, landmarks and segmentation.
- Multi-sensor/media support
- Auto-labeling functions
- ML-assisted segmentation
- Simplified data exchange
- High-threshold quality control
- Robust annotation analytics
Video Annotation
Sensor collected objects of interest can be annotated using our ML-powered tools, regardless of sequence length.
- Object tracking for large sequences
- Auto-tracking via interpolations
- Accurate instance support
- Custom class list configurations
- Additional attributes per instance
- Built-in quality control tools
Sensor Fusion
Access our best-in-class visualization and annotation infrastructure for labeling multi-sensor fusion datasets for autonomous driving.
- Auto-ground detection models
- Aggregated point cloud support
- Auto-detection of static objects
- 2D & 3D object linking and tracking
- Smart interpolation of frames
- Advanced quality check tools
AI Best Practices
The essential guide to AI training data
Discover best practices for the sourcing, labeling and analyzing of training data from TELUS Digital, a leading provider of AI data solutions.