/v1/car — brand
Send a photo, get the manufacturer back with confidence scores — 92.8% accuracy measured on hundreds of thousands of real-world photos. No logo close-up needed: the model reads the whole car.
92.8%
brand accuracy
181
brands covered
40+
Chinese & new-energy makers
Search the full list. Brands with model-level coverage show how many exact models we can name; the rest are covered by brand detection alone.
Showing 24 of 181 brands
| Brand | Models covered |
|---|---|
| Toyota | 133 |
| Nissan | 108 |
| BMW | 75 |
| Hyundai | 71 |
| Lexus | 67 |
| Chevrolet | 65 |
| Honda | 65 |
| Audi | 54 |
| Mitsubishi | 54 |
| Kia | 50 |
| Lada | 49 |
| Mazda | 49 |
| Volkswagen | 47 |
| Ford | 46 |
| Mercedes-Benz | 43 |
| Suzuki | 43 |
| Renault | 42 |
| Chery | 35 |
| BYD | 30 |
| Infiniti | 28 |
| Citroen | 26 |
| Daihatsu | 26 |
| Volvo | 25 |
| Changan | 24 |
BYD, Zeekr, Li Auto, Avatr, Xiaomi, NIO, XPeng, Denza, AITO and 30+ more — makers arriving on roads faster than legacy recognition systems can keep up.
Ferrari, Lamborghini, McLaren, Rolls-Royce, Bentley, Aston Martin, Maybach — validated on real listing photos, not press renders.
Trained on car marketplaces across North America, Europe and Asia — so Lada, GAZ, UAZ and Proton are covered alongside the global marques.
A detection model finds and crops every car in the image, so backgrounds never pollute the prediction.
A classification model reads the crop — body lines, grille, lights, proportions — and returns the top 5 brand candidates with probabilities.
no logo needed
Because the model reads the whole vehicle rather than the badge, it still works when the logo is obscured, distant or missing.
Example response
POST /v1/car?body_type=false&color=false&position=false&car_model=false
{
"success": true,
"body": {
"car_bbox": { "x1": 118, "y1": 402, "x2": 1490, "y2": 980 },
"brand": {
"label": "Porsche",
"confidence": 0.78,
"top5": [
{ "label": "Porsche", "confidence": 0.78 },
{ "label": "Audi", "confidence": 0.09 },
{ "label": "Jaguar", "confidence": 0.05 },
{ "label": "Aston Martin", "confidence": 0.03 },
{ "label": "Maserati", "confidence": 0.02 }
]
}
},
"cost": 0.02
}181 brands, spanning European, American, Japanese, Korean, Chinese, Russian and Indian manufacturers — including more than 40 Chinese and new-energy makers such as BYD, Zeekr, Li Auto, Avatr and Xiaomi.
No — it reads the entire vehicle: proportions, grille, lights and body lines. That means it works even when the badge is hidden, blurred or out of frame.
Contact us. The pipeline is retrained continuously on new marketplace data, and brand requests from customers are prioritized in the next training cycle.
A label, a confidence score, and the top-5 probability distribution — for example Porsche at 0.78 followed by the next four candidates. You choose the confidence threshold that fits your use case.
Free tier included — 150 requests per month, no credit card required. Your first car profile is one POST request away.