Automatic Photo Capture
Real-Time Visual Damage Detection

Overview & Goals

CCC Intelligent Solutions Inc. is the technology platform for the P&C insurance economy, helping to support connections between drivers, insurers, automotive manufacturers, and repair facilities. A key task that both drivers and insurers engage in is visually inspecting a damaged vehicle and capturing photos of it for both evidence documentation and determining a repair cost estimation. Under an emotionally and financially stressful situation, what if the camera itself could automatically identify damage and capture the pivotal photos one needs? What if we could remove that burden off of the user?

Problem Statement

How might we harness AI models to take photos of a damaged vehicle and provide vehicle owners an efficient and pain-free experience? 

The Current User Journey

The current method in which customers capture photos of their damaged vehicle possesses many pain points:

• The burden of taking an accurate photo (correct angle, height, and framing) is all on the user and difficult to get right without detailed guidance

• The user must take eight accurate photos all around their vehicle; a time-consuming process

• Customers expressed that they do not know what they gain from performing this long process themselves (the work to value ratio seems too high)

User Journey.png

Ideation and Preliminary Ideas

• Incorporating AR elements (and dynamic feedback in general) can move the burden of snapping an accurate photo off of the user and onto our AI system. We have ways of tracking angle of the device, their distance and position relative to the vehicle, and can identify specific parts of the vehicle in real time from the live camera view.

• Photos can be automatically taken when the AI model identifies that the user is in the right position and the correct parts of the vehicle are in view.

• An AR cube can be used to guide the user into taking particular angled close-up shots of localized damage on their vehicle.

Pre-Sketch-1.jpg
Sketches2.jpg
Sketches5.jpg
Sketches4.jpg

Early Wireframes

Early wireframes and static shot concepts were created to further articulate the UI elements on a mobile device. Specific features and components explored included:

• A top-view tracker that displays the user's position as they walk around their vehicle

• A proximity indicator that alerts the user when they are too far or too close to the vehicle.

• A dialogue on-screen that tells the user what their current step is in the process.

Sprint_4.jpg

After Effects + Adobe XD Prototype

Using After Effects, a video prototype was created to properly convey the dynamic nature of the UI and AR elements. The video prototype features many of the ideas in the initial sketches and wireframes, and also integrates more of the post-capture mobile experience. In this case, immediately after the vehicle is examined, an estimate of how much it would cost to repair the vehicle is presented to the user.

These artifacts were used for a presentation with State Farm, and thus follows some of their branding guidelines in the mobile screens.

Final_Screens_Cropped.jpg

Collaborating with Developers and Testing the Rough Product

Working with developers, a toned-back version of the application was created as a live testing prototype. Utilizing UserTesting.com, a random set of test users were asked to complete the photo capture AI to snap photos of their own vehicle.