AI vehicle damage detection for motor insurance surveys
Tap AI Damage on a survey photo, or scan every photo of a stage in one go. DigiSurvey draws a box around each damaged part and suggests the damage type, severity and whether to repair, replace or paint. You get a first draft of the parts list, and you check, correct and approve every line.
Questions first? Call 8302698307 or write to help@digitaliiisla.co.in.
From survey photos to a draft parts list in four steps
The scan works on the photos already in your survey, whether you took them with live capture, uploaded them, or received them through a customer self-survey link.
Open the photos
Go to the survey stage (Spot, Primary Survey, Under Repair or Reinspection) that has the photos you want checked.
Run the scan
Tap AI Damage on one photo, or AI damage scan — this stage to check every photo in that stage. Photos that were already scanned are skipped.
Review each box
Each damaged part is numbered on the photo, with its damage type, severity, suggested action and AI confidence. You can re-scan, hide the boxes or copy the list.
Send it to the report
Open the Damage summary, untick anything you disagree with, then use Import AI damage list in the Final report's Parts table.
One line for every damaged part it can see
Part names follow the way surveyors write them, with LH and RH taken from the driver's seat, so the list reads like your own notes and matches the workshop estimate more easily.
- Part — for example “RH headlamp”, “Front bumper”, “LH rear-view mirror”.
- Damage type — dent, scratch, crack, broken, tear, misalignment, missing or paint damage.
- Severity — minor, moderate or severe.
- Suggested action — repair, replace or paint.
- AI confidence — a percentage for each part. It is lower when the photo is dark, blurred or only shows part of the panel.
⚠ AI suggestion — surveyor verifies. Check every value against the vehicle and the photo before you use it.
One damage list for the whole survey
A front-right impact usually shows up in several photos: the front view, the 45° corner, a few close-ups. The Damage summary collects every AI-scanned photo of the survey and shows each part once, with the stage it was found in, how many photos show it and the confidence. You can untick any line you do not agree with.
Copy rows gives you tab-separated rows that you can paste into Excel. In the Final survey report, Parts → Import AI damage list adds the ticked parts as new rows. Parts that are already in the table are skipped. Each new row gets a remark such as “AI: dent, moderate, repair” and a suggested material category (metal, rubber/plastic, glass or fibre) so the depreciation can be applied. The amounts stay at zero. You enter the assessed rates and labour yourself, as you always have.
Scan results are saved with the survey. When you open a scanned photo again, its boxes load straight away and no new scan is used.
What the AI does not do
The scan only sees what is in the photo. It cannot judge hidden, mechanical or underbody damage. It cannot tell whether the damage is old or unrelated to the reported accident, and it does not decide whether a loss is payable under the policy. It does not price parts, apply depreciation rates or settle between repair and replacement for you. Those remain your professional judgement as the licensed surveyor, alongside your physical inspection and the documents.
Good photos give better suggestions: daylight, the whole panel in frame, straight-on close-ups and no heavy reflections. The GPS photo stamp camera and the stage-wise photo checklist help you get consistent photos in the first place.
Technical notes
The damage scan uses Google Gemini, called from DigiSurvey's server, so no AI key sits in your browser or phone. The DigiSurvey administrator connects the Gemini API key and chooses the model. Each user has a daily scan limit, which resets at midnight IST, and a stage scan tells you how many scans it will use before it starts. The scan needs an internet connection. You can capture photos offline and scan them after they sync. The photo is sent to Google's Gemini API to be analysed. See the Privacy Policy.
Frequently asked questions
Does the AI decide the claim amount?
No. It suggests the damaged parts, damage type, severity and whether to repair, replace or paint. When you import the list into the Final report, the amounts stay at zero. The assessment, depreciation and final figures are yours to enter and approve.
Which types of damage can it detect?
Dents, scratches, cracks, broken parts, tears, misalignment, missing parts and paint damage, each graded minor, moderate or severe. It only reports damage that is visible in the photo.
How accurate is the AI damage detection?
It depends on the photo: light, angle, distance, dirt and reflections all matter. Each finding shows an AI confidence score. Treat every line as a suggestion and check it against the vehicle and the photos.
Can I scan all the photos of a stage at once?
Yes. AI damage scan — this stage checks every photo in the open stage, skips photos already scanned, and tells you how many scans it will use from today's limit before it starts.
Is there a limit on AI scans?
Yes. Each user has a daily limit set by the DigiSurvey administrator, and it resets at midnight IST. Photos that were already scanned are not scanned again, and their saved results open instantly.
Which AI is used, and where do my photos go?
The scan uses Google Gemini through DigiSurvey's server. The photo is sent to Google's Gemini API for analysis, and the result is saved with your survey, which stays private to your account.
Related guides
Get a first-draft damage list from your next survey's photos.
Creating your account is free. Full access is just ₹7 per survey, or a one-time ₹299 for lifetime access. Payments aren't live yet — we're at launch stage, so create your account online and we'll help you get set up.