AITracker

What is AITracker?

AITracker is cloud-based software that automatically recognizes vehicle and pedestrian categories and counts them from video recordings. It is designed to simplify traffic measurement for road designers and transportation engineers by reducing the time needed for manual traffic analysis.

Instead of only producing raw counts, AITracker generates measurement outputs aligned with GDDKiA classifications (including GPR, 8+1, and EURO-6). It also provides reporting artifacts such as spreadsheets, cartograms, and daily traffic profiles that can be used for further analysis and documentation.

Key Features

• Automatic vehicle and pedestrian counting from video: Detects and counts categories directly from recorded footage, supporting traffic analysis workflows without manual frame-by-frame work.
• Up to 22 recognized object classes: Categories are specified in line with General Traffic Measurement 2025 and central traffic measurement station groupings (e.g., bicyles/scooters/cars; pedestrians, mobility aids).
• Cloud processing with scalable infrastructure: Users upload videos to the AISP cloud for processing; processing capacity scales with system load and users receive an email notification when finished.
• Multiple measurement outputs for reporting: Generates ready-made spreadsheets and diagrams, including relation-by-relation CSV, GPR 2025-compliant XML, intersection spreadsheets with pictograms, and car-by-car speed and ANPR sheets.
• Intersection and roundabout support with relation-based analysis: Can count roundabouts and intersections and measure pedestrian traffic for selected cross-sections within an intersection.
• Result handling and verification view: After processing, results are available online for up to 14 days, and users can select a counted vehicle to jump to the corresponding moment in the video for verification.

How to Use AITracker

1. Prepare your videos: In the AITracker program, configure each video directory.
2. Upload to the cloud: Upload the videos to the AISP cloud for processing.
3. Wait for processing completion: When the request is processed, you receive an email notification.
4. Download and review results: Use the online outputs (available for up to 14 days) such as spreadsheets, diagrams, and per-vehicle/relationship sheets. Video storage is not kept after the order is processed.

Use Cases

• Road and junction measurement projects: Measure traffic flow at intersections or roundabouts, then produce ready-to-share reports (e.g., daily traffic profiles and intersection cartograms).
• Pedestrian and cyclist monitoring: Count pedestrians and cyclists from video recordings, including cases where you need pedestrian intensity documentation.
• Time-bucket reporting for traffic studies: Generate measurement spreadsheets grouped by 5 minutes, 15 minutes, or 60 minutes depending on the reporting format you need.
• Data validation with video jump-to-event: Use the verification workflow where selecting a vehicle displays the moment it passes, supporting internal review and quality checks.
• Cross-section-focused intersection analysis: Filter results to a specific pedestrian crossing cross-section rather than processing the entire intersection as a single region.

FAQ

• What does AITracker produce after processing? It generates ready-made spreadsheets (including CSV and GPR 2025-compliant XML), intersection reports with pictograms, daily traffic profile charts, and per-vehicle speed/travel time and ANPR sheets.

• Does AITracker support different traffic classifications? The system supports classification aligned with GDDKiA requirements, and you can adjust vehicle classification mapping (e.g., from GPR to SCPR).

• How long are results available online? Results are available online for up to 14 days after processing.

• Are videos stored after processing? No—according to the page, AISP does not store videos after the order is processed.

• How accurate is the counting? The page states measurement accuracy figures for different tasks (e.g., traffic volume measurement and intersection measurements), including a definition for how accuracy is calculated against a manually counted test set.

Alternatives

• Manual traffic counting with video playback: A baseline approach that can be more labor-intensive and slower, but may be used when automation is not required or when specialized counting rules exceed automated categories.
• General video analytics platforms for counting objects: Tools that perform object detection/counting but may require additional configuration and custom workflows to achieve traffic-specific classifications and reporting formats.
• Other traffic measurement workflows using dedicated hardware/software: Alternatives focused on field instrumentation or station-based counting can provide traffic counts without video processing, typically fitting standardized measurement regimes.