Damage Detection

Vehicle damage from driver negligence or minor accidents is an expensive problem for rental and delivery fleets.

Today, companies perform damage inspections using paper forms or digital photography. However, photos or paper forms can be time consuming and inaccurate.

iQtransit deployed an artificial intelligence (AI) and deep learning platform to capture these damages with less human effort than paper or digital photography methods.

The iQtransit system has the capability to scan images in near real-time – without the need for human assistance. Images can be sourced from mobile apps or via fixed structures taking photos from lots, garages entrances or roadway checkpoints. The damage detection can then be used to notify clients in near real-time when damages are found.

The result: less expensive damage detection, less customer disputes and a better maintained fleet.

Sample Client: DAMAGE iD
Technology: Artificial Intelligence, Mask RCNN, Tensorflow, Pytorch
Supervised learning data requirements: 6M vehicle photos, with 50K damage photos.

Damage is automatically identified, and red circles are drawn on the original photo to show areas of concern. An API is used to communicate the damaged regions back to the reporting software system.

The iQtransit Mobile Device Autograder™ is an application of this technology. The device grades cellphones for cosmetic condition, identify scratches and cracks in a lab-like environment.

DAMAGE iD, partner and client, uses photos captured from their mobile app. Damage is then identified and used to notify rental agents or delivery fleet operations managers.