In short
Choose a sharp, well-lit front or rear three-quarter view, or a carefully cropped listing screenshot. A photo can suggest a likely car — it cannot verify one vehicle’s condition, history, or exact specification.
Choose a front or rear three-quarter view
If you have one photo, choose an angle that shows the side and either the front or rear in the same frame. Keep the roofline, wheels, grille or tail lamps visible where possible; a tight badge-only crop hides the wider shape. Vehicle-image research treats viewpoint as one of the variations that make recognition harder, because a car’s appearance changes with the camera angle.
Keep the main clues easy to see
- Use a sharp frame with even light; avoid motion blur, harsh reflections, and deep shadow over the car.
- Make the car large enough to inspect while keeping most of its outline in frame.
- Choose a view with as little obstruction from other cars, people, or objects as practical.
- If you recognize a badge or approximate year, add it as optional context. Leave guesses out.
Research on vehicle images accounts for changes in resolution, motion blur, illumination, occlusion, and viewpoint. These tips help keep visual clues visible; they do not measure Car Says Hi’s accuracy. Read Vehicle Re-Identification in Context research.
Crop a listing screenshot carefully
Remove unrelated page controls if that makes the car easier to see, but keep enough of the vehicle to judge its outline and details. If a listing gives a make, model, or year, you can enter that as context instead of relying on text that may be incomplete or wrong.
Use the result as a starting point
A photo cannot verify one car’s condition, safety, service history, ownership, originality, exact trim, or model year. Common issues mentioned in a result are research topics about a model, not a diagnosis of the pictured car. Before buying, confirm the vehicle’s identity against its records and have its condition checked in person by someone qualified.
The identity cross-check guide shows how to compare a likely identity with the car’s paperwork and official sources; the AI identification guide explains how a photo becomes a likely match and where matches can go wrong.
Sources and what they cover
Sources checked .
- Vehicle Re-Identification in Context (Kanacı, Zhu and Gong, 2018) — research on how image resolution, motion blur, illumination, occlusion and viewpoint change a vehicle’s appearance.
Spotted an error? Tell us — include the page address and what looks wrong.