Recognition - CarsXE Docs

Recognition turns vehicle images into structured identifiers you can use in the rest of the CarsXE API. Use Plate Image Recognition when the photo contains a license plate. Use VIN OCR when the photo contains a VIN plate, dashboard VIN, or sticker. Both endpoints are POST requests with JSON bodies and key as a query parameter.

When to use it

Intake workflows where users upload vehicle photos
Parking, tolling, inspections, and fleet workflows
Turning a plate photo into a plate string before plate decoding
Extracting a VIN from paperwork, dashboard stickers, or VIN plates

Plate Image Recognition

Extract a license plate number and region from a vehicle image.

Parameters

Parameter	Required	Location	Description
`key`	Yes	Query	Your CarsXE API key
`image_url`	Yes	JSON body	Public URL to an image of a vehicle/license plate

Use image_url for raw HTTP requests. Some SDKs may expose this as upload_url; the endpoint body field remains image_url.

Example

Recognize a plate

Code

curl -X POST "https://api.carsxe.com/platerecognition?key=YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"image_url":"https://imagedelivery.net/moyiiSImjJPI_EZVxNMBBw/f49aed53-d736-4370-f3f4-97418841c800/public"}'

Response

Top-level shape

{

"success": true,// always true on 200

"message": "",// empty on success

"processing_time": 117.822,// milliseconds

}

Click any object or array to expand it and see sample data. For the full interactive reference, try it live in the API Reference.

All response fields are optional. If results is empty or missing, no plate was detected.

results

Name
plate
Type
string
Description
Best plate candidate, normalized as returned by the recognition provider. Example: "9djc652".
Name
box
Type
object
Description
Bounding box around the detected plate.
Name
region
Type
object
Description
Predicted plate region with code and score.
Name
score
Type
number
Description
Confidence for the selected plate candidate, from 0 to 1.
Name
candidates
Type
array
Description
Alternate plate readings and confidence scores.
Name
dscore
Type
number
Description
Detection confidence for the plate region.
Name
vehicle
Type
object
Description
Vehicle detection details when available, including type, confidence, and vehicle bounding box.

Bounding boxes

Coordinates are pixel offsets in the original image. xmin and ymin mark the top-left corner. xmax and ymax mark the bottom-right corner. Use these values to draw overlays or crop the detected plate or vehicle.

Name
xmin
Type
number
Description
Left edge of the detected region.
Name
ymin
Type
number
Description
Top edge of the detected region.
Name
xmax
Type
number
Description
Right edge of the detected region.
Name
ymax
Type
number
Description
Bottom edge of the detected region.

Region and candidates

region - predicted registration region

The region.code identifies the predicted plate region, such as "us-ca". The region.score value is the confidence for that region prediction, from 0 to 1.

candidates - alternate plate readings

Use candidates when the top plate result is ambiguous. Each candidate includes a possible plate string and a score so you can decide whether to show alternatives, ask for review, or accept the top result automatically.

VIN OCR

Extract a 17-character VIN from an image.

Parameters

Parameter	Required	Location	Description
`key`	Yes	Query	Your CarsXE API key
`image_url`	Yes	JSON body	Public image URL or base64-encoded image string

Example

Extract a VIN

Code

curl -X POST "https://api.carsxe.com/v1/vinocr?key=YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"image_url":"https://user-images.githubusercontent.com/5663423/30922082-64edb4fa-a3a8-11e7-873e-3fbcdce8ea3a.png"}'

Response

Top-level shape

{

"success": true,// always true on 200

"vin": "JHLRD77874C026456",

"confidence": 0.9834251403808594,

}

Click any object or array to expand it and see sample data. For the full interactive reference, try it live in the API Reference.

All response fields are optional. A successful response can include one best VIN plus candidate VINs when multiple readings are possible.

VIN fields

Name
vin
Type
string
Description
Best extracted VIN candidate. Example: "JHLRD77874C026456".
Name
confidence
Type
number
Description
Confidence score between 0 and 1.
Name
box
Type
object
Description
Bounding box around the detected VIN text.
Name
candidates
Type
array
Description
Alternate VIN candidates with confidence scores and optional bounding boxes.

Image quality

Use a sharp image where the plate or VIN is readable.
Avoid glare, motion blur, extreme angles, and heavy cropping.
For VIN OCR, include the full VIN line when possible.
For base64 input on VIN OCR, send the base64 string in image_url.

Chaining Recognition With Vehicle Data

Plate Image Recognition returns a plate string; pass that into Plate Decoder when you need vehicle details. VIN OCR returns a VIN; pass that into Specifications, Market Value, Vehicle History, or Recalls.

Use the API Reference to try these endpoints live and inspect the complete schemas.

Errors

See the Errors guide for the full list.

Plate Image Recognition:

400 - missing image URL, commonly "Missing upload URL".
404 - no plate detected, commonly "No plates detected in image".
500 - recognition service error, commonly "Could not get information at this time".

VIN OCR:

400 - missing image data, invalid image data, or no text detected.
401 - API key is inactive.
404 - API key not found/disabled, or no valid VIN candidates found.
429 - usage limit exceeded.

llms.txt llms-full.txt

When to use it

Intake workflows where users upload vehicle photos
Parking, tolling, inspections, and fleet workflows
Turning a plate photo into a plate string before plate decoding
Extracting a VIN from paperwork, dashboard stickers, or VIN plates

Plate Image Recognition

Extract a license plate number and region from a vehicle image.

Parameters

Parameter	Required	Location	Description
`key`	Yes	Query	Your CarsXE API key
`image_url`	Yes	JSON body	Public URL to an image of a vehicle/license plate

Use image_url for raw HTTP requests. Some SDKs may expose this as upload_url; the endpoint body field remains image_url.

Example

Recognize a plate

Code

curl -X POST "https://api.carsxe.com/platerecognition?key=YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"image_url":"https://imagedelivery.net/moyiiSImjJPI_EZVxNMBBw/f49aed53-d736-4370-f3f4-97418841c800/public"}'

Response

Top-level shape

{

"success": true,// always true on 200

"message": "",// empty on success

"processing_time": 117.822,// milliseconds

}

Click any object or array to expand it and see sample data. For the full interactive reference, try it live in the API Reference.

All response fields are optional. If results is empty or missing, no plate was detected.

results

Name
plate
Type
string
Description
Best plate candidate, normalized as returned by the recognition provider. Example: "9djc652".
Name
box
Type
object
Description
Bounding box around the detected plate.
Name
region
Type
object
Description
Predicted plate region with code and score.
Name
score
Type
number
Description
Confidence for the selected plate candidate, from 0 to 1.
Name
candidates
Type
array
Description
Alternate plate readings and confidence scores.
Name
dscore
Type
number
Description
Detection confidence for the plate region.
Name
vehicle
Type
object
Description
Vehicle detection details when available, including type, confidence, and vehicle bounding box.

Bounding boxes

Name
xmin
Type
number
Description
Left edge of the detected region.
Name
ymin
Type
number
Description
Top edge of the detected region.
Name
xmax
Type
number
Description
Right edge of the detected region.
Name
ymax
Type
number
Description
Bottom edge of the detected region.

Region and candidates

region - predicted registration region

The region.code identifies the predicted plate region, such as "us-ca". The region.score value is the confidence for that region prediction, from 0 to 1.

candidates - alternate plate readings

VIN OCR

Extract a 17-character VIN from an image.

Parameters

Parameter	Required	Location	Description
`key`	Yes	Query	Your CarsXE API key
`image_url`	Yes	JSON body	Public image URL or base64-encoded image string

Example

Extract a VIN

Code

curl -X POST "https://api.carsxe.com/v1/vinocr?key=YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"image_url":"https://user-images.githubusercontent.com/5663423/30922082-64edb4fa-a3a8-11e7-873e-3fbcdce8ea3a.png"}'

Response

Top-level shape

{

"success": true,// always true on 200

"vin": "JHLRD77874C026456",

"confidence": 0.9834251403808594,

}

Click any object or array to expand it and see sample data. For the full interactive reference, try it live in the API Reference.

All response fields are optional. A successful response can include one best VIN plus candidate VINs when multiple readings are possible.

VIN fields

Name
vin
Type
string
Description
Best extracted VIN candidate. Example: "JHLRD77874C026456".
Name
confidence
Type
number
Description
Confidence score between 0 and 1.
Name
box
Type
object
Description
Bounding box around the detected VIN text.
Name
candidates
Type
array
Description
Alternate VIN candidates with confidence scores and optional bounding boxes.

Image quality

Use a sharp image where the plate or VIN is readable.
Avoid glare, motion blur, extreme angles, and heavy cropping.
For VIN OCR, include the full VIN line when possible.
For base64 input on VIN OCR, send the base64 string in image_url.

Chaining Recognition With Vehicle Data

Use the API Reference to try these endpoints live and inspect the complete schemas.

Errors

See the Errors guide for the full list.

Plate Image Recognition:

400 - missing image URL, commonly "Missing upload URL".
404 - no plate detected, commonly "No plates detected in image".
500 - recognition service error, commonly "Could not get information at this time".

VIN OCR:

400 - missing image data, invalid image data, or no text detected.
401 - API key is inactive.
404 - API key not found/disabled, or no valid VIN candidates found.
429 - usage limit exceeded.

llms.txt llms-full.txt