AUTOMAVIS builds deep learning systems for automated vehicle inspection — damage detection, severity scoring, and part-level defect mapping, deployed on whatever capture hardware you already have.
Detections are drawn as precise polygons and linked to the part they sit on.
Our Damage Inspector generates detailed polygons around each damage and uniquely identifies it across multiple views.
Read automatically for fleet tagging, with plate visibility controlled by your privacy settings:
The plate ties every scan to a specific vehicle across a fleet — no separate tagging step.
Matches this scan to the vehicle's last one by plate, so the report only shows what's new.
Where required, the plate and the window interiors are both blurred before the image reaches cloud storage.
The same detection pipeline runs underneath — only the capture hardware changes to fit what you already have.
See it in action →
Your cameras. Our AI. Results via API.
Portable. No fixed cameras required.
Walk around. AR-guided. Nothing missed.
Underbody inspection using mobile phone cameras — checking for structural damage, leakage, corrosion, and more.
AMV's core Deep Learning and Computer Vision models.
Trained on more than 100,000 annotated images to recognize the damage types shown above.
AMV's inspection app, coordinating capture across every mounted phone and feeding it to the web acquisition dashboard.
The handheld capture mode above runs DIL+, identifying parts as the inspector moves around the vehicle.
Every finding DIL produces is available the same way, whatever sits on the other end.
{ "part": "front_left_door", "type": "dent", "confidence": 0.942, "severity": "moderate" }
Every polygon carries a type, a confidence score, and the part it was mapped to.
{ "thresholds": { "dent": 0.75, "scratch": 0.7, "rim_damage": 0.8, "deformation": 0.75 } }
Each damage category has its own threshold, tunable per customer to balance precision and recall.
The same API plugs into your existing ERP or fleet system — request a vehicle's status, get its damage report back.
{ "vehicle_id": "RC-4471", "event": "return" }
{ "images": [ "https://.../8841/front.jpg", "https://.../8841/left.jpg", "https://.../8841/rear.jpg", "https://.../8841/right.jpg" ], "new_findings": 2, "damages": [ { "part": "rear_bumper", "type": "scratch", "severity": "minor", "image": "https://.../8841/rear.jpg" }, { "part": "front_wheel", "type": "rim_damage", "severity": "severe", "image": "https://.../8841/right.jpg" } ] }
Whichever of the three capture modes you use, every result lands in the same web-based dashboard.
AUTOMAVIS is a spinoff from CERTH/ITI — one of Greece's leading research institutes in computer vision and AI.
We work with fleet operators, dealerships, and manufacturers to deploy AMV on whichever scanning mode fits their site.
Tell us a bit about your inspection line and we'll get back to you.
