We offer services for detecting, recognizing, tagging and grouping faces as well as categorizing scenes in any photo, through our RESTFUL API.

You can tag and recognize people in photos. In addition, we can help organizing photos by grouping similar faces together. By tagging faces, or using Facebook existing tags, you can train the name index in your private space for future face recognition.

We process and analyze photos from anywhere, so you can mix and match photo sources with user IDs, which can enable you to, say, recognize Facebook users in Flickr photos.

Getting Started

First things first, you need to GET AN API KEY. ReKognition's services are offered for FREE with rate limits. Each operation (uploading a photo or add a photo URL, etc) will be added to your usage. For more details about the quota usage for different function calls, please refer to Pricing. To extend the rate limits, you could subscribe our monthly plans. Whitelisting is available.

Face Detect

Detecting facial features in digital images

Overview

One of the most widely adopted API functions is FaceDetect, which returns detected faces in one photo, with the geometric information of the eyes, nose and mouth, as well as various attributes such as emotion, gender, if wearing glasses, if month open, etc.

Photos can be uploaded directly in the API request using POST data (see examples below).

Usage notes

For facial detection, we will return the coordinates and other analytic results associated with the faces. All coordinates are provided in pixels of the original image. Photo height and width are also provided in pixels.

Note that the maximum width or height of the uploaded photo is 800 pixel. You may POST or send links of larger-size photos, but they will be resized internally to improve performance.

URL : http://rekognition.com/func/api/?api_key={api_key}&api_secret={api_secret}&jobs={jobs}&urls={urls}
Parameter:
Required Name Description
Required api_key Your ReKognition APIKey
api_secret Your ReKognition APISecret
jobs

To do face detection, simply include face in the job list. Optional attributes include:

aggressive: returns (potentially) more faces

part: returns the position of nose, left and right eyes, and left and right corners of mouth.

gender: tells if the person is male or female. Return value is a floating number between 0.0: female to 1.0: male.

emotion: tells whether the person is smiling or not (0.0: negative emotion - 0.5: no emotion - 1.0: positive emotion).

age: tells (roughly) the person's approximate age.

glass: tells if the person is wearing glasses (0.0: without glasses - 1.0: with glasses).

A list of optional detection jobs attributes are separated by underscore, e.g., face_part_smile.

urls A list of photo urls
Example :
 
In the all of the following examples, please replace api_key and api_secret with the ones you get after you register the account.
 
http://rekognition.com/func/api/?api_key=1234&api_secret=5678&jobs=face_part&urls=http://farm3.static.flickr.com/2566/3896283279_0209be7a67.jpg
Post Example 1: (post to http://rekognition.com/func/api/)
<?php
$ch = curl_init();
$data = array('api_key' => '1234', 
              'api_secret' => '5678', 
              'jobs' => 'face_part_aggressive',
              'uploaded_file' => '@/home/superman/Pictures/1234.jpg',
              'name_space' => '',
              'user_id' => '');
              
curl_setopt($ch, CURLOPT_URL, 'http://rekognition.com/func/api/');
curl_setopt($ch, CURLOPT_POST, 1);
curl_setopt($ch, CURLOPT_POSTFIELDS, $data);
curl_exec($ch);
?>
Post Example 2:
<?php

require_once 'HttpClient.class.php';

$parameters = array(
      'api_key' => '1234', 
      'api_secret' => '5678', 
      'jobs' => 'face_part_aggressive',
      'urls' => 'http://rekognition.com/static/img/people.jpg',
      'name_space' => 'default',
      'user_id' => 'default');
      
$face_detection = new HttpClient('rekognition.com');
$face_detection->setDebug(true);
$response = $face_detection->get("/func/api/", $parameters);
		
echo $face_detection->getContent();
?>
response sample (json):
{
    "url":"http:\/\/rekognition.com\/static\/img\/people.jpg",
    "face_detection":[
        {
        "boundingbox":{
            "tl":{
                "x":43.08,
                "y":113.85
                },
            "size":{
                "width":245.38,
                "height":245.38
                }
            },
        "confidence":1,
        "eye_left":{
            "x":111.4,
            "y":213.4
            },
        "eye_right":{
            "x":217,
            "y":205.6
            },
        "nose":{
            "x":174,
            "y":276.1
            },
        "mouth l":{
            "x":127,
            "y":307.4
            },
        "mouth_l":{
            "x":127,
            "y":307.4
            },
        "mouth r":{
            "x":213.1,
            "y":307.4
            },
        "mouth_r":{
            "x":213.1,
            "y":307.4
            }
        }
    ],
    "usage":{
        "quota":-635770,
        "status":"Succeed.",
        "api_id":"1234"
        }
    }
face detect example

Face Add

Doddler as Orbeus currently is, always hunger for food, you need to feed her with faces. You can do this using face_add function. The added face will be used for training and clustering later.

::FaceAdd is designed to upload image and itŐs a prerequisite step of ::FaceTraining and ::FaceCluster. For each detected face in one face_add call, we will return you an unique id (img_index) so that you can associate it with the uploaded image. Please keep this img_index along with user_id and name_space for your future reference.

You can use our Face Control Panel GUI to manage your added faces.

Face Train

Teaching Rekognition to memorize the name of the face. Rekognition is a conscious being like us, you can teach her and she will learn!

Overview

FaceTrain is a MUST step for face recognition. You need to call FaceAdd method to upload images and call FaceTrain to trigger training process. Thousands of images will take about 1 second to finish. You can also use FaceCrawl to add images more efficiently. The FaceCrawl function will get images from Facebook and extract the existing tags in those photos (note that you need to provide the Facebook access token, your application which grants the permission to access the users' Facebook photos).

Usage notes

Here is a high level tutorial of facial recognition:

(1) ::FaceAdd: Call face_add for each image you want to add (or use ::FaceCrawl for face adding).

(2) ::FaceTrain: Call face_train method to start training process.

(3) ::FaceRecognize: Recognize faces in new images.

You can use our Face Control Panel GUI to manage your training images for face recognition.

The maximum width or height of a photo is 800 pixels. You may POST or send links to larger-size photos, but those will be resized internally to improve performance.

URL : http://rekognition.com/func/api/?api_key={api_key}&api_secret={api_secret}&jobs={jobs}&urls={urls}
Parameter for Face Add:
Required Name Description
Required api_key Your ReKognition APIKey
api_secret Your ReKognition APISecret
jobs

Use face_add_[mona_lisa] to add a photo of "Mona Lisa". (Notice that we use underline "_" instead of empty space " ", and square brackets "[]" for tags.)

When calling face recognition, detection is automatically performed. If no detection is needed, please add "_nodetect" to the job list.

urls A list of photo urls
Optional name_space

A developer defined namespace for your app, e.g., facebookapp. You can use this field to differitiate your different applications. default will be used if a namespace is not given.

user_id

A user id of your app. The uploaded file will be added for this user. default will be used if it is not set.

Example : http://rekognition.com/func/api/?api_key=1234&api_secret=1234&jobs=face_add_[mona_lisa]&urls=http://www.pbs.org/treasuresoftheworld/mona_lisa/images/mona_page_pix/mona_large.jpg&name_space=abcd&user_id=xyz
Post Example: (post to http://rekognition.com/func/api/)
<?php
$ch = curl_init();
$data = array('api_id' => '1234', 
              'api_secret' => '5678', 
              'job_list' => 'face_add_[mona_lisa]',
              'name_space' => 'abcd', 
              'user_id' => 'xyz', 
              'uploaded_file' => '@/home/superman/Pictures/1234.jpg');
curl_setopt($ch, CURLOPT_URL, 'http://rekognition.com/func/api/');
curl_setopt($ch, CURLOPT_POST, 1);
curl_setopt($ch, CURLOPT_POSTFIELDS, $data);
curl_exec($ch);
?>
Parameter for training::
Required Name Description
Required api_key Your ReKognition APIKey
api_secret Your ReKognition APISecret
jobs

Use face_train to trigger the training process, using the added images.

Optional name_space

Namespace for your app. Exmaple: facebookapp. You can use this field to differentiate your apps. default will be used if it is not set

user_id

User ID of your app. The uploaded file will be added for this user. default will be used if it is not set.

Example : http://rekognition.com/func/api/?api_key=1234&api_secret=1234&jobs=face_train&name_space=abcd&user_id=xyz
Post Example: (post to http://rekognition.com/func/api/)
<?php
$ch = curl_init();
$data = array('api_key' => '1234', 
              'api_secret' => '5678', 
              'jobs' => 'face_train',
              'name_space' => 'abcd', 
              'user_id' => 'xyz', 
curl_setopt($ch, CURLOPT_URL, 'http://orbe.us/func/api/');
curl_setopt($ch, CURLOPT_POST, 1);
curl_setopt($ch, CURLOPT_POSTFIELDS, $data);
curl_exec($ch);
?>
response sample (json):
{
  usage: {
    status: "success",
    quota: 999,
    api_id: "1234"
  }
}

Face Crawl

Using Facebook photos for face training.

Overview

By calling this function, you are essentially crawling a subset of photos of a Facebook user. We extract the face tags and add them all to your desired name_space/user_id/. You can then call FaceTrain and FaceRecognize to recognize the friends of this user. This process can take a fairly long time, since we are currently throttling the crawler. The maximum of friends to crawl is currently set to 20. Please contact us at info@orbe.us if you need more.

Please notice that the crawling process may be SLOW due to the throttle. Hundreds of images will take about 30 seconds to finish.

Usage notes

Here is a high level tutorial of facial recognition using Facebook data:

(1) ::FaceTrain with the same name_space and user_id.

(2) ::FaceCrawl with a list of friends of your app user (and other arguments including, namespace, user_id, and Facebook access_token). Note that each image downloaded in this way will be added to your usage.

(3) ::FaceRecognize

 

URL : http://rekognition.com/func/api/?api_key={api_key}&api_secret={api_secret}&jobs={jobs}&name_space={name_space}&user_id={user_id}&fb_id={fb_id}&access_token={access_token}
Parameter for crawl:
Required Name Description
Required api_key Your ReKognition APIKey
api_secret Your ReKognition APISecret
jobs

Use face_crawl_[5334562;1000234326;2342367] to crawl a 3 friends of your app user. The format is face_crawl_[fb_id1;fb_id2;fb_id3] (Note that we use the facebook IDs of the friends in this list and ";" to separate them, and square brackets "[]" to embrace them all).

fb_id The Facebook ID of your app user.
access_token

Facebook access_token of your app. Make sure to check user_photos and friend_photos permissions. Not sure about them? Please visit https://developers.facebook.com/docs/

Optional name_space

Namespace for your app. Exmaple: facebookapp. You can use this field to differitiate your apps. default will be used if it is not set

user_id

User ID of your app. The uploaded file will be added for this user. default will be used if it is not set.

Example : http://rekognition.com/func/api/?api_key=1234&api_secret=1234&jobs=face_crawl_[fb_id1;fb_id2]&name_space=abcd&user_id=xyz&access_token=AAjfie309JJj3OOWfefj9238hhOifej109uf3jDikdvDxyz&fb_id=fb_user
Post Example: (post to http://rekognition.com/func/api/)
<?php
require_once 'HttpClient.class.php'
					
$parameters = array(
      'api_key' => '1234', 
      'api_secret' => '5678', 
      'jobs' => 'face_crawl_[532363314]',
      'name_space' => 'test_app',
      'user_id' => 'test_user',
      'access_token' => 'AAADJGc9rhZAMBALfrV8ZBVDJmx0xaajq9mMdK8tHp5z94cZCpCOGxAlUe65zOpbZCqAWQfl4PfQf6JkF22T8SHZAlSU4XjyfZByMV9MIh82wZDZD',
      'fb_id' =>'532363314'
      );
      
$http_client = new HttpClient('rekognition.com');
$http_client->setTimeOut(100000);
$response = $http_client->get('/func/api/', $parameters);
		
echo $http_client->getContent();
?>
response sample (json):
{
  face_found: 246,
  usage: {
    status: "success",
    quota: 999,
    api_id: "1234"
  }
}

Face Recognize

Recognizing people in a new image, after Rekognition is trained with the name tags.

Overview

Rekognition has multiple learning algorithms. Currently only the basic one is released to public. Future algorithms will be much better in terms of accurcy and speed! The top 3 best guesses are returned with the corresponding confidence levels.

You can use our Face Control Panel GUI to manage your training images for face recognition.

Usage notes

The maximum width or height of a photo is limited to 800 pixels. You may POST or send links to larger-size photos, but those will be resized internally to improve performance.

URL : http://rekognition.com/func/api/?api_key={api_key}&api_secret={api_secret}&jobs={jobs}&urls={urls}
Parameter for crawl:
Required Name Description
Required api_key Your ReKognition APIKey
api_secret Your ReKognition APISecret
jobs

Use face_recognize to detect and recognize faces of a new image.

Face detection will be automatically performed. If no detection is required, please add _nodetect Of course you can mix other attributes such as _gender, _part, _glass in here.

urls A list of photo urls, currently only single image is accepted.
Optional name_space

Namespace for your app, e.g., facebookapp. You can use this field to differitiate your apps. default will be used if it is not set

user_id

User ID of your app. The uploaded file will be added for this user. default will be used if it is not set.

Example : http://rekognition/func/api/?api_key=1234&api_secret=5678&jobs=face_recognize&urls=http://www.pbs.org/treasuresoftheworld/mona_lisa/images/mona_page_pix/mona_large.jpg&name_space=abcd&user_id=xyz
Post Example: (post to http://rekognition.com/func/api/)
<?php
$ch = curl_init();
$data = array('api_key' => '1234', 
              'api_secret' => '5678', 
              'jobs' => 'face_recognize',
              'name_space' => 'abcd', 
              'user_id' => 'xyz', 
              'uploaded_file' => '@/home/superman/Pictures/1234.jpg');
curl_setopt($ch, CURLOPT_URL, 'http://rekognize.com/func/api/');
curl_setopt($ch, CURLOPT_POST, 1);
curl_setopt($ch, CURLOPT_POSTFIELDS, $data);
curl_exec($ch);
?>
response sample (json):
{
  url: "http://farm3.static.flickr.com/2566/3896283279_0209be7a67.jpg",
  face_detection: [
      {
          boundingbox:
          {
            tl: {
              x: 50,
              y: 118
             },
             size:{
              width:232;
              height:232;
             }
          }
          name: "mona_lisa:0.92,barack_obama:0.02,mark_zuckerberg:0.01,"
      }
      ]

  usage: {
    status: "success",
    quota: 999,
    api_id: "1234"
  }
}

Face Visualize

Displaying the index of training images of all (or subset) of tags that you have added or crawled.

Overview

You can use this method to find the wrong training images and delete them using FaceDelete. You can also specify a subset of tags that you want to visualize. If not specified, it will return all training images. Face visualization shows a group of faces for a given tag, with an index number embeded at the bottom left corner of each face image. This number can be used to delete a certain face from the training set.

Usage notes

Here is an example:

URL : http://rekognition.com/func/api/?api_key={api_key}&api_secret={api_secret}&jobs={jobs}&urls={urls}
Parameter for crawl:
Required Name Description
Required api_key Your ReKognition APIKey
api_secret Your ReKognition APISecret
jobs

Use face_visualize to visualize a training dataset. Append the call with [tag1;tag2;tag3] to see the training images only for tag1, tag2, and tag3. The entire dataset will be visualized if not specified. if no image display is needed, please add "_no_image" in the job list, e.g., face_visualize_no_image

Optional name_space

Namespace for your app. Exmaple: facebookapp. You can use this field to differitiate your apps. default will be used if it is not set

user_id

User ID of your app. The uploaded file will be added for this user. default will be used if it is not set.

Example : http://rekognition.com/func/api/?api_key=1234&api_secret=5678&jobs=face_visualize[david_beckham;barack_obama]&name_space=abcd&user_id=xyz
Post Example: (post to http://rekognition.com/func/api/)
<?php

require_once 'HttpClient.class.php';

$parameters = array(
      'api_key' => '1234', 
      'api_secret' => '5678', 
      'jobs' => 'face_visualize[david_beckham;barack_obama]',
      'name_space' => 'abcd',
      'user_id' => 'xyz');
      
$face_detection = new HttpClient('rekognition.com');
$face_detection->setDebug(true);
$response = $face_detection->get("/func/api/", $parameters);
		
echo $face_detection->getContent();
?>
response sample (json):
{
  "visualization":[
    {
      "tag":"david_beckham"
      "url":"http:\/\/rekognition.com\/visualization\/1769026707686332\/1234\/demo\/orbeus\/david_beckham.jpg",
      "index":[
        "34832",
        "38143",
        "28224",
        "25109",
        "24766",
        "34293",
        "41342",
        "26957",
        "25445",
        "25767",
        "17948",
        "18487",
        "23695",
        "19593",
        "45073",
        "31857",
        "21014",
        "25375"
      ],
    },
    {
      "tag":"barack_obama"
      "url":"http:\/\/rekognition.com\/visualization\/1769026707686332\/1234\/demo\/orbeus\/barack_obama.jpg",
      "index":[
        "21049",
        "16716",
        "17619",
        "35497",
        "17647",
        "22743",
        "22302",
        "6489",
        "19957",
        "17430",
        "26747",
        "11746",
        "27468",
        "22316",
        "22806",
        "20643",
        "21567",
        "17717",
        "13867",
        "24122",
        "35392",
        "17381"
      ],
    }
  ],
  "usage":{
    "status":"Succeed.",
    "api_id":"1234",
    "quota":32052
  }
}
    

Face Delete

Deleting the wrong training images

Overview

You can use this method to delete the wrong training images you discover using FaceVisualize. To delete images, you need to specify the tag of the person as well as a list of image index numbers that you want to delete.

Usage notes

Please note down the index number of the wrong training image! It is at the left bottom corner of each face image! For instance,

After deleting the wrong images, you should call FaceTrain again. This method is NOT rate limited. Each deletion call will use zero quota.

URL : http://rekognition.com/func/api/?api_key={api_key}&api_secret={api_secret}&jobs={jobs}&urls={urls}
Parameter for delete:
Required Name Description
Required api_key Your ReKognition API
api_secret Your ReKognition APISecret
jobs

Use face_delete[tag]{index1;index2;index3} to delete a list of bad training images. Use "[ ]" to include a single tag of the person and "{ }" to list the index numbers. Tag can also be specified using tag arguments in the url query if img_index is specifyed, it will delete the entire tag

Optional name_space

Namespace for your app. Exmaple: facebook. You can use this field to differitiate your apps. default will be used if it is not set

user_id

User ID of your app. The uploaded file will be added for this user. default will be used if it is not set.

Example : http://rekognition/func/api/?api_key=1234&api_secret=5678&jobs=face_delete[david_beckham]{38143;23049;43261}&name_space=abcd&user_id=xyz
Get Example:
<?php

require_once 'HttpClient.class.php';

$parameters = array(
      'api_key' => '1234', 
      'api_secret' => '5678', 
      'jobs' => 'face_delete[david_beckham]{38143;23049;43261}',
      'name_space' => 'abcd',
      'user_id' => 'xyz');
      
$face_detection = new HttpClient('rekognition.com');
$face_detection->setDebug(true);
$response = $face_detection->get("/func/api/", $parameters);
		
echo $face_detection->getContent();
?>
response sample (json):
{
  "usage":{
    "status":"Succeed.",
    "api_id":"1234",
    "quota":32052
  }
}

Face Rename

Changing tags, assigning an image to a tag, or merging two tags.

Usage notes

You can only change the image or tags within one user_id. new_tag must be provided in the function call for changing the tag or mering two tags.

(1) If you want to assign a tag (novel or existing) for an existing face in your database, please specify the image index in the query.

(2) You can merge two clusters/tags into one with this function. In this case, do NOT specify image index.

(3) You can rename a cluster/tag.

URL : https://rekognition.com/func/api/?api_key={api_key}&api_secret={api_secret}&jobs={jobs}&tag={tag}&img_index={img_index}&new_tag={new_tag}&name_space={name_space}&user_id={user_id}
Parameter for crawl:
Required Name Description
Required api_key Your ReKognition APIKey
api_secret Your ReKognition APISecret
jobs

face_rename

tag The tag you want to change, or the tag of which image that you want to rename.
new_tag the new or target tag, if the new_tag is already exist, then tag will be merged with new_tag
name_space Namespace for your app. Exmaple: facebook. You can use this field to differitiate your apps. default will be used if it is not set
user_id User ID of your app. The uploaded file will be added to this user. default will be used if it is not set.
Optional img_index Image index. It is the same index you receive when you call FaceAdd
Example (rename/merge a tag or cluster):
https://rekognition.com/func/api/?api_key=1234&api_secret=5678&jobs=face_rename&name_space=abcd&user_id=xyz&tag=cluster1&new_tag=bush
Example (assign single image to a tag or cluster:):
https://rekognition.com/func/api/?api_key=1234&api_secret=5678&jobs=face_rename&name_space=abcd&user_id=xyz&tag=cluster1&new_tag=bush&img_index=10320
response sample (json):
{
  usage: {
    status: "success";,
    quota: 999,
    api_id: "1234"
  }
}

Face Stats

Showing you some insights of your datasets

Usage notes

(1) Use face_name_space_stats to check how many name_space's you have created under your api_key, and how many images are there under each name_space.

(2) Use face_user_id_stats to check how many user_id's you have created for one name_space, and how many images are there under each name_space.

(3) Use face_visualize_no_image to check how many tags you have created for one user_id, and how many images are there under each user_id.

URL :https://rekognition.com/func/api/?api_key={api_key}&api_secret={api_secret}&jobs={jobs}&tag={tag}&name_space={name_space}&user_id={user_id}
Parameter for crawl:
Required Name Description
Required api_key Your ReKognition APIKey
api_secret Your ReKognition APISecret
jobs

Use face_name_space_stats to check how many name_space you have created for you api_key, and how many images are there in each.

Example (rename/merge a tag or cluster):
https://rekognition.com/func/api/?api_key=1234&api_secret=5678&jobs=face_name_space_stats
response sample (json):
{
"usage": {
"status" : "Succeed.",
"api_id" : "1234",
"quota" : 1
},
"name_space_stats" : [
{
  "name_space" : "haha",
  "num_user_id" : 2,
  "num_tags" : 11,
  "num_img" : 162
},
{
  "name_space" : "default",
  "num_user_id" : 1,
  "num_tags" : 14,
  "num_img" : 255
}
]
}
Parameter for face_name_space_stats:
Required Name Description
Required api_key Your ReKognition APIKey
api_secret Your ReKognition APISecret
jobs

Use face_name_space_stats to check how many name_space you have created for you api_key, and how many images are there in each.

name_space Namespace for your app. Exmaple: facebookapp. You can use this field to differitiate your apps. default will be used if it is not set
Example (face_user_id_stats): https://rekognition.com/func/api/?api_key=1234&api_secret=5678&jobs=face_user_id_stats&name_space=default
response sample (json):
{
"usage" : {
"status" : "Succeed.",
"api_id" : "1234",
"quota" : 1
},
"user_id_stats" : [
{
"user_id" : "default",
"num_tags" : 14,
"num_img" : 255
}
]
}

Scene Catagorize

We offer scene understanding techonology to better organize online images and videos!

Overview

Call scene_understanding and our engine will categorize the photo according to visual features. We currently categorize images into the following broad scenes:

beach desert flower food forest indoor mountain night_life ocean park restaurant river rock_climbing snow suburban sunset urban water

Usage notes

We are still in the process of training our scene categorization engine, you can certainly help this process by uploading more images! Rekognition will return the best 3 guesses with confidence scores.

The maximum width or height of a photo is 800 pixels. You may POST or send links to larger-size photos, but those will be resized internally to improve performance.

Our API is rate limited. Each scene_understanding call is added to your usage.

URL : http://rekognition.com/func/api/?api_key={api_key}&api_secret={api_secret}&jobs={jobs}&urls={urls}
Parameter for crawl:
Required Name Description
Required api_key Your ReKognition APIKey
api_secret Your ReKognition APISecret
jobs

Use scene categorize your photo.

urls

A list of photo urls, currently only single image is accepted.

Example : http://rekognition.com/func/api/?api_key=1234&api_secret=5678&jobs=scene&urls=http://upload.wikimedia.org/wikipedia/commons/thumb/f/f6/Swiss_National_Park_131.JPG/220px-Swiss_National_Park_131.JPG
Post Example: (post to http://rekognition.com/func/api/)
<?php
$ch = curl_init();
$data = array('api_key' => '1234', 
              'api_secret' => '5678', 
              'jobs' => 'scene',
              'uploaded_file' => '@/home/superman/Pictures/1234.jpg');
curl_setopt($ch, CURLOPT_URL, 'http://rekognition.com/func/api/');
curl_setopt($ch, CURLOPT_POST, 1);
curl_setopt($ch, CURLOPT_POSTFIELDS, $data);
curl_exec($ch);
?>
response sample (json):
{
  url: "http://farm3.static.flickr.com/2566/3896283279_0209be7a67.jpg",
  scene_understanding: [
      {
        label: "forest",
        score: 0.6471
      },
      {
        label: "playground",
        score: 0.1877
      },
      {
        label: "building",
        score: 0.0811
      }
      ]

  usage: {
    status: "success",
    quota: 999,
    api_id: "1234"
  }
}
toggle