Skip to main content

Snapshot API Reference

Welcome to the Plate Recognizer Snapshot API! You can use our API to access our API endpoints, which can read license plates from images.

For detailed instructions on how to use Snapshot Cloud, go here.

For detailed instructions on how to install the Snapshot SDK, go here.

We have multiple language bindings. You can view code examples in various languages when you scroll down.

Other use cases:

Snapshot Cloud API​

Authentication​

Plate Recognizer Snapshot Cloud is only available to registered users. Sign up for a Free Trial and get an API Token. It has to be included in all API calls. The HTTP headers must contain:

Authorization: Token YOUR_API_TOKEN

danger

For the commands below, make sure to replace YOUR_API_TOKEN with your API Token. For example, Authorization: Token 3a0effff73919f898b69ac65a32dc12347769564.

Get your token from, here.

Read Number Plates from an Image​

This Snapshot API endpoint reads all license plates from an image.

tip
  • By default, we only return vehicles with a license plate. To include vehicles without a license plate, set the engine configuration to {"detection_mode":"vehicle"}.
  • To programmatically process a bunch of images in a folder, refer to our Snapshot Guides.

HTTP Request​

POST https://api.platerecognizer.com/v1/plate-reader/

The CORS policy of this endpoint allows requests from all origins.

POST Parameters​

ParameterRequiredDescription
uploadYesThe file to be uploaded. The parameter can either be the file bytes (using Content-Type multipart/form-data) OR a base64 encoded image. This parameter becomes optional if upload_url parameter is present.
upload_urlNoThe url of file to be uploaded. This parameter is to be used as an alternative to upload parameter.
regionsNoMatch the license plate pattern of specific state(s) or country(ies). Best practice is to indicate the top 3-4 states/countries of vehicle plates that camera will be observing. This optimizes the engine to correctly decode zero vs letter O, one vs letter I, etc. Note that even if a foreign vehicle arrives, engine will still decode that foreign vehicle correctly. Accepts multiple arguments or comma-separated values. See list of states and countries.
camera_idNoUnique camera identifier.
timestampNoISO 8601 timestamp. For example, 2019-08-19T13:11:25. The timestamp has to be in UTC.
mmcNoPredict vehicle make, model, orientation and color. This feature is available for an additional fee. Set parameter to true (mmc=true) if you have this feature enabled/purchased to get vehicle make, model and color. Possible values are true or false.
configNoAdditional engine configuration. See details.
#On Linux
#Calling the API with 2 regions (Mexico and California).
curl -o /tmp/car.jpg https://app.platerecognizer.com/static/demo.jpg # Get an image
curl -F "upload=@/tmp/car.jpg" \
-F regions=`mx` \
-F regions=us-ca \
-H "Authorization: Token YOUR_API_TOKEN" \
https://api.platerecognizer.com/v1/plate-reader/

#Calling the API with a custom engine configuration and 2 regions (Mexico and California).
curl -F "upload=@/path/to/car.jpg" \
-F regions=mx,us-ca \
-F config="{\"region\":\"strict\"}" \
-H "Authorization: Token YOUR_API_TOKEN" \
https://api.platerecognizer.com/v1/plate-reader/

#Calling the API with an image URL.
curl -X POST -F upload_url="https://app.platerecognizer.com/static/demo.jpg" -H "Authorization: Token YOUR_API_TOKEN" https://api.platerecognizer.com/v1/plate-reader
#On Windows
#Calling the API with 2 regions (Mexico and California).
curl -o car.jpg https://app.platerecognizer.com/static/demo.jpg
curl -F "[email protected]" ^
-F regions=`mx` ^
-F regions='us-ca' ^
-H "Authorization: Token YOUR_API_TOKEN" ^
https://api.platerecognizer.com/v1/plate-reader/

#Calling the API with a custom engine configuration and region California.
curl -F "upload=@c:\path\to\car.jpg" ^
-F regions='us-ca' ^
-F config="{\"region\":\"strict\"}" ^
-H "Authorization: Token YOUR_API_TOKEN" ^
https://api.platerecognizer.com/v1/plate-reader/

Response​

The response is a list of all the license plates found in the image.

{
"processing_time": 288.758,
"results": [
{
"box": {
"xmin": 143,
"ymin": 481,
"xmax": 282,
"ymax": 575
},
"plate": "nhk552",
"region": {
"code": "gb",
"score": 0.747
},
"vehicle": {
"score": 0.798,
"type": "Sedan",
"box": {
"xmin": 67,
"ymin": 113,
"xmax": 908,
"ymax": 653
}
},
"score": 0.904,
"candidates": [
{
"score": 0.904,
"plate": "nhk552"
}
],
"dscore": 0.99,
// Make Model, Orientation and Color are only available if you set mmc=true
"model_make": [
{
"make": "Riley",
"model": "RMF",
"score": 0.306
}
],
"color": [
{
"color": "black",
"score": 0.937
}
],
"orientation": [
{
"orientation": "Front",
"score": 0.937
}
]
}
],
"filename": "1617_7M83K_car.jpg",
"version": 1,
"camera_id": null,
"timestamp": "2020-10-12T16:17:27.574008Z"
}

Each license plate has the following elements:

AttributeDescription
results/plateText of the license plate.
results/boxBounding box for the license plate. Coordinates in pixels of the top-left and bottom-right corners of the plate.
results/dscoreConfidence level for plate detection. See below for more details.
results/scoreConfidence level for reading the license plate text. See below for more details.
results/vehicle/typeVehicle type: Big Truck, Bus, Motorcycle, Pickup Truck, Sedan, SUV, Van, Unknown.
results/vehicle/scoreConfidence level for vehicle type prediction. If we cannot find a vehicle, the score is set to 0.
results/vehicle/boxVehicle bounding box. If we cannot find a vehicle, the coordinates are all 0.
results/region/codeRegion of license plate. Returns a code from the country list.
results/region/scoreConfidence level for license plate region.
results/candidatesList of predictions for the license plate value. The first element is the top prediction (same as results/plate).
results/model_make/makePrediction of vehicle make.
results/model_make/modelPrediction of vehicle model.
results/model_make/scoreConfidence level for vehicle make and model prediction.
results/color/colorVehicle color. One of black, blue, brown, green, red, silver, white, yellow, unknown.
results/color/scoreConfidence level for vehicle color prediction.
results/orientation/orientationVehicle orientation. One of Front, Rear, Unknown.
results/orientation/scoreConfidence level for vehicle orientation prediction.
  • Scores vary between 0 and 1. A prediction with a score close to 1 indicates high confidence.

  • The value dscore is the confidence level of the object detection. A score close to 1 indicates a high confidence. For example, if you don't want to miss any vehicles and you can tolerate some false positives, you can decide to use all results with a score above 0.3. However, if you want to avoid false positives and can tolerate missing a few vehicles, then discarding results below 0.7 is preferable. It depends on the use case and quality of the images you use.

info

View complete examples for ALPR API integration. Easily do batching and use the API on a video. Examples are written in multiple languages.

Error Responses

Engine Configuration​

A lookup request can accept the parameter config. It is a JSON value to change the engine configuration. It can include the following values:

  • {"detection_rule":"strict"}: The license plates that are detected outside a vehicle will be discarded.
  • {"detection_mode":"vehicle"}: The default detection mode (plate) only returns vehicles if a license plate is also detected. To get vehicles without plates, use the value vehicle. See the response format.
  • {"region":"strict"}: Only accept the results that exactly match the templates of the specified region. For example, if the license plate of a region is 3 letters and 3 numbers, the value abc1234 will be discarded. For regions with vanity license plates (e.g. in us-ca), we do not recommend the use of Strict Mode. Otherwise, the engine will discard the vanity plates. A region must be specified in order for this parameter to work.
  • {"threshold_d":0.2, "threshold_o":0.6}: By default the engine will use those thresholds to filter the detection and OCR results. Anything below that will be discarded. You can set different values.
  • {"mode":"fast"}: The number of detection steps is always 1. On average it gives a 30% speed-up. May result in lower accuracy when using images with small vehicles.
  • {"mode":"redaction"}: Used for Plate Recognizer Blur. It includes more candidates during the plate detection step. This configuration will miss fewer plates but will increase the number of false positives (objects that are not license plates).
  • {"text_formats":["..."]}: List of regular expressions. They are used to guide the engine's predictions. For example, if you know the output is generally 3 digits and 2 letters, you would use - {"text_formats":["[0-9][0-9][0-9][a-z][a-z]"]}. More information about the format. The engine may not follow the format when the predictions are too different. If you want to exactly match the regexp, you should also use "region":"strict" (see above).
  • {"plates_per_vehicle":1}: Maximum number of license plates that a vehicle is expected to have.
  • {"zoom_in_vehicles":4}: Maximum number of vehicles Snapshot performs a zoom on to improve plate recognition. Increasing this parameter could help with high resolution images, but may have a negative effect on latency. New in version 1.47.0.

Examples​

Here are a couple of examples of how to use the config parameter (with Shell or Python).

#On Linux
#Custom threshold and fast mode:
curl -F "upload=@/path/to/car.jpg" \
-F config="{\"mode\":\"fast\", \"threshold_d\":0.2, \"threshold_o\":0.6}" \
-H "Authorization: Token YOUR_API_TOKEN" \
https://api.platerecognizer.com/v1/plate-reader/

#Strict region matching:
curl -F "upload=@/path/to/car.jpg" \
-F config="{\"region\":\"strict\"}" \
-F region=us-ca \
-H "Authorization: Token YOUR_API_TOKEN" \
https://api.platerecognizer.com/v1/plate-reader/

#Prediction must include a vehicle:
curl -F "upload=@/path/to/car.jpg" \
-F config="{\"detection_rule\":\"strict\"}" \
-H "Authorization: Token YOUR_API_TOKEN" \
https://api.platerecognizer.com/v1/plate-reader/
#On Windows
#Custom threshold and fast mode:
curl -F "upload=@c:\path\to\car.jpg" ^
-F config="{\"mode\":\"fast\", \"threshold_d\":0.2, \"threshold_o\":0.6}" ^
-H "Authorization: Token YOUR_API_TOKEN" ^
https://api.platerecognizer.com/v1/plate-reader/

#Strict region matching:
curl -F "upload=@c:\path\to\car.jpg" ^
-F config="{\"region\":\"strict\"}" ^
-F region=us-ca ^
-H "Authorization: Token YOUR_API_TOKEN" ^
https://api.platerecognizer.com/v1/plate-reader/

#Prediction must include a vehicle:
curl -F "upload=@c:\path\to\car.jpg" ^
-F config="{\"detection_rule\":\"strict\"}" ^
-H "Authorization: Token YOUR_API_TOKEN" ^
https://api.platerecognizer.com/v1/plate-reader/

Statistics​

Get number of recognition calls done during the current month. This API endpoint is for Snapshot Cloud only. For Snapshot SDK, use the info API.

HTTP Request​

GET https://api.platerecognizer.com/v1/statistics/

curl -H "Authorization: Token YOUR_API_TOKEN" \
https://api.platerecognizer.com/v1/statistics/

JSON Response​

{
"usage": {
"month": 1,
"calls": 128,
"year": 2019,
"resets_on": "2019-01-14T00:00:00Z"
},
"total_calls": 2500
}

Response fields description:

AttributeDescription
callsNumber of API calls made during the current period.
monthMonth of the current period.
yearYear of the current period.
resets_onDate when the counter will reset.
total_callsMaximum number of API calls you can make during the period. Need more? Upgrade.

Errors​

Error CodeMeaning
403Forbidden: you do not have enough credits to perform this request or your API Token is wrong.
413Payload Too Large response status code indicates that the request entity is larger than limits defined by our server. See upload limits.
429Indicates the user has sent too many requests in a given amount of time. The Free Trial Snapshot API Cloud Plan has a limit of 1 lookup per second. A Snapshot API Cloud Subscription has a limit of 8 lookups per second. The Snapshot SDK does not have any lookup limits per second. Subscribe for a higher number of calls per second for your API Cloud plan. The response is {"detail":"Request was throttled. Expected available in 1 second.","status_code":429}.

Snapshot On-Premise API​

Our service is also available on-premises. Get started with the SDK. It has a similar interface as the recognition API and it is hosted locally. See differences below.

info

Are you using the on-premise API from a different device? Make sure that your firewall allows it. For example, you can disable the Windows Defender firewall to check if it is blocking the API calls.

Recognition API​

HTTP Request​

POST http://localhost:8080/v1/plate-reader/

The CORS policy of this endpoint allows requests from all origins. This endpoint has no maximum calls per second contrary to our Cloud API.

Parameters​

All the parameters from recognition API.

#On Linux
#Calling the API with just the image
curl -o car.jpg https://app.platerecognizer.com/static/demo.jpg ## Get an image
curl -F "[email protected]" \
http://localhost:8080/v1/plate-reader/

#Sending the image in base64
curl -F "upload=$(base64 'image.jpg')" http://localhost:8080/v1/plate-reader/

#Calling with API with optional parameters config and mmc
#The region is set to Mexico and California
curl -F "upload=@/path/to/car.jpg" \
-F regions=`mx` \
-F regions=us-ca \
-F mmc=true \
-F config="{\"mode\":\"fast\"}" \
http://localhost:8080/v1/plate-reader/

JSON Response​

{
"usage": {
"max_calls": 1000,
"calls": 44
},
"results": [
// Same as cloud API example above.
],
"camera_id": "null",
"timestamp": "2020-01-16 17:00:00",
"filename": "car.jpg"
}

SDK Version​

Show version, webhooks and usage information.

HTTP Request​

GET http://localhost:8080/info/

JSON Response​

Returns the SDK version, license key and webhooks.

{
"version": "1.3.8",
"license_key": "XXX",
"total_calls": 2500,
"usage": {"calls": 10},
"webhooks": []
}

WebHooks​

Our service also supports webhooks. It allows you to receive a HTTP POST request to a target URL of your choosing. Go to webhooks settings page to manage your webhooks or add a new webhook target.

  • The target_url should return a valid HTTP status code (200). If the target_url consistently returns an error code, the hook will be removed and an email will be sent to the account owner.
  • To quickly test out this feature, you can use http://webhook.site. It generates a unique target URL and displays all the requests made to that URL.
  • A webhook is sent for each call to the recognition API even when no license plate is found.

Example of POST payload​

{
"hook": {
"target": "http://localhost:8081/",
"id": 2,
"event": "image.done"
},
"data": {
"processing_time": 0.186,
"timestamp": "2019-08-31T14:22:06.983Z",
"results": [
{
"box": {
"xmin": 563,
"ymin": 530,
"ymax": 579,
"xmax": 702
},
"plate": "765410",
"score": 0.9131534678711563,
"dscore": 0.4084282568529709
},
{
"box": {
"xmin": 563,
"ymin": 530,
"ymax": 579,
"xmax": 702
},
"plate": "830265",
"score": 0.7700640306685624,
"dscore": 0.8493143507578083
}
],
"filename": "14_22_test.jpeg",
"version": 1,
"camera_id": null
}
}

Data Only Webhook​

Snapshot: HTTP Request​

POST target_url

The request body contains the payload above in JSON format.

Webhook With Image​

Snapshot: HTTP Request​

POST target_url

The request body contains a multipart/form-data content. Receiving this data is like receiving the content of a form with an input of type file. It has two fields:

  • upload contains the image binary content
  • json is the JSON encoded data. See the example of the payload above.
info

The bounding box of the license plate are calculated based on the image sent by webhook.

Receiving Webhook Data​

Do you need help creating the webhook receiver? Get started with our examples.

Detection Zones​

Detection Zones exclude overlay texts, street signs or other objects.

Countries​

Check the countries codes for regions parameter.

Floating button icon
CTRL + .