Identifies vehicle type (sedan, SUV, truck, bus, motorcycle), make, color, and direction of travel alongside license plate reading. Search every vehicle detection by appearance.
Visylix Vehicle Classifier is a dedicated AI model that runs alongside ANPR on every vehicle detection. It classifies vehicle type across 10+ categories (sedan, SUV, hatchback, pickup, truck, bus, motorcycle, bicycle, van, trailer), estimates vehicle color across 11 common classes, detects make where externally visible, and tracks general direction of travel. Classification attributes attach directly to each detection record so forensic search can filter by appearance without requiring a readable plate. This is critical for getaway-vehicle searches, fleet tracking, parking analytics, or high-volume toll and checkpoint scenarios where plates are often dirty, damaged, or occluded.
Core capabilities of the Vehicle Classifier model.
Distinguishes 10+ vehicle types: sedan, SUV, hatchback, pickup, truck, bus, motorcycle, bicycle, van, and trailer. 95%+ accuracy on well-positioned cameras.
11 common color classes (white, black, red, blue, silver, grey, green, yellow, orange, brown, gold) with confidence scores per detection.
Tracks approximate direction (northbound/southbound/eastbound/westbound) using frame-over-frame displacement, useful for wrong-way detection and flow analytics.
Every ANPR license plate detection carries full classification attributes. Every classification carries the plate when visible. No separate pipeline, no latency penalty.
Search for "red SUV between 2pm and 4pm" or "white truck westbound on Camera 3" without needing a plate. Critical for getaway vehicles, damaged plates, and dirty-plate scenarios.
Per-camera adaptive thresholds refine accuracy over 30 days as the model learns angle, lighting, and perspective specifics of each camera location.
Type classification works reliably at night with IR-equipped cameras. Color detection degrades gracefully in low light with confidence-scored outputs.
Under 20ms per frame on GPU inference, running concurrently with ANPR, motion, and tracking without added hardware cost.
Real-world applications for Vehicle Classifier.
Search for suspect vehicles by type and color even when the plate is unreadable, damaged, or not yet identified. Reduce investigation time from millions of detections to a small candidate set.
Track which vehicle types use which lots, identify commercial vehicles in passenger-only zones, and enforce policy with automatic alerts on unauthorized vehicle types.
Classify every vehicle for accurate counting and billing where class-based pricing applies. Identify commercial fleet vehicles, oversized loads, and motorcycles that need different handling.
Match company-owned fleet appearance (white trucks with blue livery, for example) to verify presence, detect unauthorized after-hours use, and correlate with GPS telematics.
Classify traffic composition by type and color across intersections and arterials for infrastructure planning, congestion modeling, and emissions policy enforcement.
Performance and deployment details.
Add Vehicle Classifier to your video pipeline in minutes.
Assign the model to specific cameras with zone definitions and sensitivity settings through the web UI or API.
The model processes video frames in real time, generating structured detection events with bounding boxes and metadata.
Receive instant alerts via webhooks, trigger automated workflows, or query detections through the REST API.
Explore other computer vision capabilities.
Talk to our team to see this model in action on your video feeds.