0

I am using the upper mentioned library (Google Cloud Vision Client Library v1) in PHP to assign labels to images... so far so good. It all works, except it returns fewer results than on the google test page... as far as I understand it has to do with a "max_results" parameter which defaults to 10, but I am not able to find where/how to set it manually... There was a similar question here on Python and there it was as simple as passing it as a parameter - I have tried many options to do this in PHP, but apparently I am doing something wrong...

Here is a link to the documentation : https://googleapis.github.io/google-cloud-php/#/docs/cloud-vision/v0.19.3/vision/v1/imageannotatorclient?method=labelDetection I am guessing I have to pass it to the "optionalArgs" parameter... but not exactly sure how to do this...

Here is more or less what my code is:

require __DIR__ . '/vendor/autoload.php';

use Google\Cloud\Vision\V1\ImageAnnotatorClient;

$this->client = new ImageAnnotatorClient();
$response = $this->client->labelDetection(...THE IMAGE...);
$labels = $response->getLabelAnnotations();

if ($labels) {
    foreach ($labels as $label) {
        // do something with $label->getDescription()
    }
}

Anyone got an idea how to get more results in the $labels array?

Виктор Иванов
  • 377
  • 1
  • 12

2 Answers2

2

New Method

Since the other answer I provided seems to be deprecated, I am going to provide a sample that uses the setMaxResults method on the Feature object.

$imageAnnotatorClient = new ImageAnnotatorClient();
$gcsImageUri = 'some/image.jpg';
$source = new ImageSource();
$source->setGcsImageUri($gcsImageUri);
$image = new Image();
$image->setSource($source);
$type = Feature_Type::FACE_DETECTION;
$featuresElement = new Feature();
$featuresElement->setType($type);
$featuresElement->setMaxResults(100); // SET MAX RESULTS HERE
$features = [$featuresElement];
$requestsElement = new AnnotateImageRequest();
$requestsElement->setImage($image);
$requestsElement->setFeatures($features);
$requests = [$requestsElement];
$imageAnnotatorClient->batchAnnotateImages($requests);

Deprecated Method

The maxResults value gets specified in the Image constructor

An example of this code can be found in the source code for the Image object.

 $imageResource = fopen(__DIR__ . '/assets/family-photo.jpg', 'r');
 $image = new Image($imageResource, [
   'FACE_DETECTION',
   'LOGO_DETECTION'
 ], [
     'maxResults' => [
         'FACE_DETECTION' => 1
     ],
     'imageContext' => [
          ....
     ]
   ]
]);
craigts
  • 2,857
  • 1
  • 25
  • 25
  • Hello craigts, Thank you for the reply. I have seen this in the docs and dependency files, but this just creates an object that "Represents an image to be annotated".... I can successfully do that... but how do I actually get the labels I need from here? If I pass this object to the ImageAnnotatorClient-> labelDetection() it doesn't work... as far as I understand you are not supposed to do that. So any ideas? – Виктор Иванов Mar 26 '19 at 09:39
0

OK, so for anybody who still may need this here is a working example

    use Google\Cloud\Vision\Image;
    use Google\Cloud\Vision\VisionClient;

    $imageResource = fopen(__DIR__ .'/'. $fileIMG, 'r');

    $thePic = new Image($imageResource, [
         'LABEL_DETECTION',
          'LOGO_DETECTION',
          'TEXT_DETECTION'
     ], [
         'maxResults' => [
             'LABEL_DETECTION' => 20,
             'LOGO_DETECTION' => 20,
             'TEXT_DETECTION' => 20
         ],
         'imageContext' => []
     ]);


    $vision = new VisionClient();
    $result = $vision->annotate($thePic);

    $finalLabels = array();

    // do the same for $results->text(), $results->logos()
    if($result->labels()){
        foreach ($result->labels() as $key => $annonObj) {
            $tmp = $annonObj->info();
            $finalLabels[] = $tmp['description'];
        }
    }

But... as it says in the official documentation

  • Please note this client will be deprecated in our next release. In order
  • to receive the latest features and updates, please take
  • the time to familiarize yourself with {@see Google\Cloud\Vision\V1\ImageAnnotatorClient}.

So I still need a way to do this using the ImageAnnotatorClient class... any ideas anyone?

Виктор Иванов
  • 377
  • 1
  • 12