Video Query Service (VQS) is TruVideo's core intelligence engine. It transforms unstructured video into structured, searchable, actionable data — damage types, severity scores, timestamps, text, objects, and conditions — all queryable via API.
Video arrives from any source — SDK, web upload, Send2Me link, API batch. VQS normalizes format, resolution, and encoding automatically.
AI models analyze every frame: object detection, damage classification, severity scoring, OCR, scene segmentation, and temporal mapping. Running on BYOAI architecture — swap models without rewriting pipelines.
Results are stored as structured records with full provenance — timestamp, source frame, confidence score, model version. Query via API, stream via webhooks, or aggregate in dashboards.
Scratches, dents, cracks, corrosion, fluid leaks — classified by type, severity, and location. Trained on 80M+ real-world inspection videos across multiple asset types.
Automated condition grades for assets, components, and surfaces. Consistent, repeatable, and calibrated against human expert assessments.
Not just what damage exists — when it appeared. VQS builds asset memory by comparing inspections over time, creating verifiable condition timelines.
OCR for VIN plates, serial numbers, work orders, labels, and signage. Structured extraction feeds directly into DMS, ERP, and CMMS systems.