Cloudinary Blog

Facial attribute detection with Microsoft's Face API

Facial Attribute Detection with Microsoft's Face API

Many of the photos displayed on the internet these days are of people. If your website or mobile application displays photos that include people, you will want to make sure that their faces are included in the delivered images when cropping and manipulating them to fit your graphic design and responsive layout. You may even want to further manipulate an image according to the faces present, for example, adding a harlequin mask overlay on all of their eyes, where each mask is adjusted to the correct size and orientation (although not a typical use case, it's a cool example of using advanced facial attribute detection):

Ruby:
cl_image_tag("cloudinary_team.jpg", :transformation=>[
  {:width=>700, :radius=>"max", :crop=>"scale"},
  {:flags=>"region_relative", :gravity=>"adv_eyes", :overlay=>"harlequinmask", :width=>1.7}
  ])
PHP:
cl_image_tag("cloudinary_team.jpg", array("transformation"=>array(
  array("width"=>700, "radius"=>"max", "crop"=>"scale"),
  array("flags"=>"region_relative", "gravity"=>"adv_eyes", "overlay"=>"harlequinmask", "width"=>1.7)
  )))
Python:
CloudinaryImage("cloudinary_team.jpg").image(transformation=[
  {'width': 700, 'radius': "max", 'crop': "scale"},
  {'flags': "region_relative", 'gravity': "adv_eyes", 'overlay': "harlequinmask", 'width': 1.7}
  ])
Node.js:
cloudinary.image("cloudinary_team.jpg", {transformation: [
  {width: 700, radius: "max", crop: "scale"},
  {flags: "region_relative", gravity: "adv_eyes", overlay: "harlequinmask", width: "1.7"}
  ]})
Java:
cloudinary.url().transformation(new Transformation()
  .width(700).radius("max").crop("scale").chain()
  .flags("region_relative").gravity("adv_eyes").overlay(new Layer().publicId("harlequinmask")).width(1.7)).imageTag("cloudinary_team.jpg");
JS:
cloudinary.imageTag('cloudinary_team.jpg', {transformation: [
  {width: 700, radius: "max", crop: "scale"},
  {flags: "region_relative", gravity: "adv_eyes", overlay: new cloudinary.Layer().publicId("harlequinmask"), width: "1.7"}
  ]}).toHtml();
jQuery:
$.cloudinary.image("cloudinary_team.jpg", {transformation: [
  {width: 700, radius: "max", crop: "scale"},
  {flags: "region_relative", gravity: "adv_eyes", overlay: new cloudinary.Layer().publicId("harlequinmask"), width: "1.7"}
  ]})
React:
<Image publicId="cloudinary_team.jpg" >
  <Transformation width="700" radius="max" crop="scale" />
  <Transformation flags="region_relative" gravity="adv_eyes" overlay="harlequinmask" width="1.7" />
</Image>
Angular:
<cl-image public-id="cloudinary_team.jpg" >
  <cl-transformation width="700" radius="max" crop="scale">
  </cl-transformation>
  <cl-transformation flags="region_relative" gravity="adv_eyes" overlay="harlequinmask" width="1.7">
  </cl-transformation>
</cl-image>
.Net:
cloudinary.Api.UrlImgUp.Transform(new Transformation()
  .Width(700).Radius("max").Crop("scale").Chain()
  .Flags("region_relative").Gravity("adv_eyes").Overlay(new Layer().PublicId("harlequinmask")).Width(1.7)).BuildImageTag("cloudinary_team.jpg")
Android:
MediaManager.get().url().transformation(new Transformation()
  .width(700).radius("max").crop("scale").chain()
  .flags("region_relative").gravity("adv_eyes").overlay(new Layer().publicId("harlequinmask")).width(1.7)).generate("cloudinary_team.jpg");
iOS:
imageView.cldSetImage(cloudinary.createUrl().setTransformation(CLDTransformation()
  .setWidth(700).setRadius("max").setCrop("scale").chain()
  .setFlags("region_relative").setGravity("adv_eyes").setOverlay("harlequinmask").setWidth(1.7)).generate("cloudinary_team.jpg")!, cloudinary: cloudinary)
Cloudinary team with masks

Face Detection is a great feature that enables the automatic modification of images according to the detected faces within an image, making it simple to intelligently crop, position, resize and transform your images appropriately.

At Cloudinary we strive to create an enriched environment that can solve our customers media asset related needs. By taking a holistic approach to image management, we create partnerships with leading companies developing image processing and media related technologies that can extend our internal features and capabilities and cater customers with more “complex” needs. Our add-ons feature services pre-integrated into Cloudinary that are tailored for developing, extending, and operating web and mobile apps.

We have recently partnered with Microsoft's Cognitive Services which provides a Face API for high precision face detection with state-of-the-art cloud-based algorithms. The Face API technology is fully integrated within our Advanced Facial Attributes Detection add-on that can do more than just detect the human faces in an image. The Advanced Facial Attribute Detection add-on can also extract meaningful advanced data about the face(s) in an image, including the exact location of facial features. This allows you even greater control over your image categorization, and to automatically use these details to smartly crop, position, rotate and overlay images based on the detected facial features.

Microsoft Cognitive Services logo

How to automatically detect facial attributes

Cloudinary supports uploading images using a cloud-based API. You can request further information while uploading the image by setting the detection parameter to adv_face when calling Cloudinary's upload API and Advanced Facial Attribute Detection is utilized to automatically extract detailed face attributes from the uploaded image. The detected faces are returned in the JSON response with rectangles (left, top, width and height) indicating the location of faces in the image in pixels, the exact position details of the eyes, mouth, eyebrows, nose and lips, as well as a series of face related attributes from each face such as pose, gender and age. See the Advanced Facial Attribute Detection documentation for more information. The code sample below uploads the lady image while requesting that the facial attributes are also returned in the JSON response:

Ruby:
Cloudinary::Uploader.upload("lady.jpg", 
              :detection: "adv_face")
PHP:
\Cloudinary\Uploader::upload("lady.jpg", 
              array(
               "detection": "adv_face"));
Python:
cloudinary.uploader.upload("lady.jpg", 
              detection = "adv_face")
Node.js:
cloudinary.uploader.upload("lady.jpg", 
              function(result) {console.log(result); }, { detection: "adv_face" });
Java:
cloudinary.uploader().upload("lady.jpg", 
              Cloudinary.asMap("detection", "adv_face"));

Ruby:
cl_image_tag("lady.jpg")
PHP:
cl_image_tag("lady.jpg")
Python:
CloudinaryImage("lady.jpg").image()
Node.js:
cloudinary.image("lady.jpg")
Java:
cloudinary.url().imageTag("lady.jpg");
JS:
cloudinary.imageTag('lady.jpg').toHtml();
jQuery:
$.cloudinary.image("lady.jpg")
React:
<Image publicId="lady.jpg" >

</Image>
Angular:
<cl-image public-id="lady.jpg" >

</cl-image>
.Net:
cloudinary.Api.UrlImgUp.BuildImageTag("lady.jpg")
Android:
MediaManager.get().url().generate("lady.jpg");
iOS:
imageView.cldSetImage(cloudinary.createUrl().generate("lady.jpg")!, cloudinary: cloudinary)
Original uploaded image

The example JSON snippet below contains the results of the upload response when applying advanced facial attribute detection on the uploaded image. The response includes very detailed information regarding the face that was automatically detected in the image above:

{
...
 "info": 
  {"detection": 
    {"adv_face": 
      {"status": "complete",
       "data": 
        [{"bounding_box": 
           {"top": 234.0, "left": 216.0, "width": 244.0, "height": 244.0},
          "attributes": 
           {"smile": 0.649,
            "head_pose": {"pitch": 0.0, "roll": -6.7, "yaw": 0.6},
            "gender": "female",
            "age": 30.3,
            "facial_hair": {"moustache": 0.0, "beard": 0.0, 
                            "sideburns": 0.0}},
          "facial_landmarks": 
           {"mouth": 
             {"left": {"x": 277.1, "y": 410.6},
              "right": {"x": 410.2, "y": 395.1},
              "under_lip": 
               {"bottom": {"x": 352.8, "y": 445.0},
                "top": {"x": 349.1, "y": 431.8}},
              "upper_lip": 
               {"bottom": {"x": 346.3, "y": 415.1},
                "top": {"x": 345.4, "y": 407.5}}},
            "eyebrow": 
             {"left_outer": {"x": 224.1, "y": 298.0},
              "left_inner": {"x": 306.6, "y": 283.4},
              "right_inner": {"x": 361.4, "y": 279.8},
              "right_outer": {"x": 428.8, "y": 272.2}},
            "eye": 
             {"left_outer": {"x": 258.5, "y": 314.1},
              "left_top": {"x": 277.0, "y": 306.2},
              "left_bottom": {"x": 277.1, "y": 315.6},
              "left_inner": {"x": 296.4, "y": 312.9},
              "right_inner": {"x": 373.4, "y": 305.2},
              "right_top": {"x": 388.0, "y": 294.5},
              "right_bottom": {"x": 389.0, "y": 306.0},
              "right_outer": {"x": 406.5, "y": 300.2},
              "left_pupil": {"x": 278.3, "y": 309.4},
              "right_pupil": {"x": 386.0, "y": 298.7}},
            "nose": 
             {"tip": {"x": 341.6, "y": 381.6},
              "root_left": {"x": 321.9, "y": 314.6},
              "root_right": {"x": 343.6, "y": 311.8},
              "left_alar_top": {"x": 312.7, "y": 359.7},
              "right_alar_top": {"x": 359.2, "y": 351.2},
              "left_alar_out_tip": {"x": 305.4, "y": 380.4},
              "right_alar_out_tip": {"x": 374.3, "y": 367.5}}}}]}}},
}

Dynamic image manipulation with facial attribute detection

Cloudinary supports on-the-fly image manipulation using simple parameters in HTTP delivery URLs. Based on the position of facial attributes detected by the Advanced Facial Attribute Detection add-on, Cloudinary can crop your images to focus on the detected facial features, while providing a large set of image transformation and cropping options when using a Cloudinary delivery URL. To focus an automatic crop on the detected faces, simply set the crop parameter to thumb, fill or crop and the gravity parameter to adv_faces (set gravity to adv_face for focusing on the single largest detected face in the image). The resulting images are dynamically generated on-the-fly and the result is delivered via a fast CDN.

For example, to deliver a 300x300 thumbnail of the lady image shown above:

Ruby:
cl_image_tag("lady.jpg", :gravity=>"adv_face", :height=>300, :width=>300, :crop=>"thumb")
PHP:
cl_image_tag("lady.jpg", array("gravity"=>"adv_face", "height"=>300, "width"=>300, "crop"=>"thumb"))
Python:
CloudinaryImage("lady.jpg").image(gravity="adv_face", height=300, width=300, crop="thumb")
Node.js:
cloudinary.image("lady.jpg", {gravity: "adv_face", height: 300, width: 300, crop: "thumb"})
Java:
cloudinary.url().transformation(new Transformation().gravity("adv_face").height(300).width(300).crop("thumb")).imageTag("lady.jpg");
JS:
cloudinary.imageTag('lady.jpg', {gravity: "adv_face", height: 300, width: 300, crop: "thumb"}).toHtml();
jQuery:
$.cloudinary.image("lady.jpg", {gravity: "adv_face", height: 300, width: 300, crop: "thumb"})
React:
<Image publicId="lady.jpg" >
  <Transformation gravity="adv_face" height="300" width="300" crop="thumb" />
</Image>
Angular:
<cl-image public-id="lady.jpg" >
  <cl-transformation gravity="adv_face" height="300" width="300" crop="thumb">
  </cl-transformation>
</cl-image>
.Net:
cloudinary.Api.UrlImgUp.Transform(new Transformation().Gravity("adv_face").Height(300).Width(300).Crop("thumb")).BuildImageTag("lady.jpg")
Android:
MediaManager.get().url().transformation(new Transformation().gravity("adv_face").height(300).width(300).crop("thumb")).generate("lady.jpg");
iOS:
imageView.cldSetImage(cloudinary.createUrl().setTransformation(CLDTransformation().setGravity("adv_face").setHeight(300).setWidth(300).setCrop("thumb")).generate("lady.jpg")!, cloudinary: cloudinary)
150x150 thumbnail of lady.jpg

Cloudinary can also dynamically crop your images based on the position of detected eyes. Simply set the gravity parameter to adv_eyes (g_adv_eyes for URLs) to center the image on the detected eyes. The example below delivers a 200x60 thumbnail centered on the eyes:

Ruby:
cl_image_tag("lady.jpg", :gravity=>"adv_eyes", :width=>200, :height=>60, :crop=>"thumb")
PHP:
cl_image_tag("lady.jpg", array("gravity"=>"adv_eyes", "width"=>200, "height"=>60, "crop"=>"thumb"))
Python:
CloudinaryImage("lady.jpg").image(gravity="adv_eyes", width=200, height=60, crop="thumb")
Node.js:
cloudinary.image("lady.jpg", {gravity: "adv_eyes", width: 200, height: 60, crop: "thumb"})
Java:
cloudinary.url().transformation(new Transformation().gravity("adv_eyes").width(200).height(60).crop("thumb")).imageTag("lady.jpg");
JS:
cloudinary.imageTag('lady.jpg', {gravity: "adv_eyes", width: 200, height: 60, crop: "thumb"}).toHtml();
jQuery:
$.cloudinary.image("lady.jpg", {gravity: "adv_eyes", width: 200, height: 60, crop: "thumb"})
React:
<Image publicId="lady.jpg" >
  <Transformation gravity="adv_eyes" width="200" height="60" crop="thumb" />
</Image>
Angular:
<cl-image public-id="lady.jpg" >
  <cl-transformation gravity="adv_eyes" width="200" height="60" crop="thumb">
  </cl-transformation>
</cl-image>
.Net:
cloudinary.Api.UrlImgUp.Transform(new Transformation().Gravity("adv_eyes").Width(200).Height(60).Crop("thumb")).BuildImageTag("lady.jpg")
Android:
MediaManager.get().url().transformation(new Transformation().gravity("adv_eyes").width(200).height(60).crop("thumb")).generate("lady.jpg");
iOS:
imageView.cldSetImage(cloudinary.createUrl().setTransformation(CLDTransformation().setGravity("adv_eyes").setWidth(200).setHeight(60).setCrop("thumb")).generate("lady.jpg")!, cloudinary: cloudinary)
200x60 thumbnail centered on eyes

Thanks to the detailed information on the position of facial attributes detected by the Advanced Facial Attribute Detection add-on, Cloudinary can add overlays while taking into account the pose of the face, and automatically scale and rotate the overlay accordingly.

Ruby:
cl_image_tag("HarlequinMask.jpg", :width=>150, :crop=>"scale")
PHP:
cl_image_tag("HarlequinMask.jpg", array("width"=>150, "crop"=>"scale"))
Python:
CloudinaryImage("HarlequinMask.jpg").image(width=150, crop="scale")
Node.js:
cloudinary.image("HarlequinMask.jpg", {width: 150, crop: "scale"})
Java:
cloudinary.url().transformation(new Transformation().width(150).crop("scale")).imageTag("HarlequinMask.jpg");
JS:
cloudinary.imageTag('HarlequinMask.jpg', {width: 150, crop: "scale"}).toHtml();
jQuery:
$.cloudinary.image("HarlequinMask.jpg", {width: 150, crop: "scale"})
React:
<Image publicId="HarlequinMask.jpg" >
  <Transformation width="150" crop="scale" />
</Image>
Angular:
<cl-image public-id="HarlequinMask.jpg" >
  <cl-transformation width="150" crop="scale">
  </cl-transformation>
</cl-image>
.Net:
cloudinary.Api.UrlImgUp.Transform(new Transformation().Width(150).Crop("scale")).BuildImageTag("HarlequinMask.jpg")
Android:
MediaManager.get().url().transformation(new Transformation().width(150).crop("scale")).generate("HarlequinMask.jpg");
iOS:
imageView.cldSetImage(cloudinary.createUrl().setTransformation(CLDTransformation().setWidth(150).setCrop("scale")).generate("HarlequinMask.jpg")!, cloudinary: cloudinary)
Harlequin mask

For example, in order to automatically overlay the above image of a harlequin mask scaled to 170% relative to the detected eyes in the main image:

Ruby:
cl_image_tag("lady.jpg", :flags=>"region_relative", :gravity=>"adv_eyes", :overlay=>"HarlequinMask", :width=>1.7, :crop=>"scale")
PHP:
cl_image_tag("lady.jpg", array("flags"=>"region_relative", "gravity"=>"adv_eyes", "overlay"=>"HarlequinMask", "width"=>1.7, "crop"=>"scale"))
Python:
CloudinaryImage("lady.jpg").image(flags="region_relative", gravity="adv_eyes", overlay="HarlequinMask", width=1.7, crop="scale")
Node.js:
cloudinary.image("lady.jpg", {flags: "region_relative", gravity: "adv_eyes", overlay: "HarlequinMask", width: "1.7", crop: "scale"})
Java:
cloudinary.url().transformation(new Transformation().flags("region_relative").gravity("adv_eyes").overlay(new Layer().publicId("HarlequinMask")).width(1.7).crop("scale")).imageTag("lady.jpg");
JS:
cloudinary.imageTag('lady.jpg', {flags: "region_relative", gravity: "adv_eyes", overlay: new cloudinary.Layer().publicId("HarlequinMask"), width: "1.7", crop: "scale"}).toHtml();
jQuery:
$.cloudinary.image("lady.jpg", {flags: "region_relative", gravity: "adv_eyes", overlay: new cloudinary.Layer().publicId("HarlequinMask"), width: "1.7", crop: "scale"})
React:
<Image publicId="lady.jpg" >
  <Transformation flags="region_relative" gravity="adv_eyes" overlay="HarlequinMask" width="1.7" crop="scale" />
</Image>
Angular:
<cl-image public-id="lady.jpg" >
  <cl-transformation flags="region_relative" gravity="adv_eyes" overlay="HarlequinMask" width="1.7" crop="scale">
  </cl-transformation>
</cl-image>
.Net:
cloudinary.Api.UrlImgUp.Transform(new Transformation().Flags("region_relative").Gravity("adv_eyes").Overlay(new Layer().PublicId("HarlequinMask")).Width(1.7).Crop("scale")).BuildImageTag("lady.jpg")
Android:
MediaManager.get().url().transformation(new Transformation().flags("region_relative").gravity("adv_eyes").overlay(new Layer().publicId("HarlequinMask")).width(1.7).crop("scale")).generate("lady.jpg");
iOS:
imageView.cldSetImage(cloudinary.createUrl().setTransformation(CLDTransformation().setFlags("region_relative").setGravity("adv_eyes").setOverlay("HarlequinMask").setWidth(1.7).setCrop("scale")).generate("lady.jpg")!, cloudinary: cloudinary)
Harlequin masked face

See the Advanced Facial Attribute Detection documentation for more information on the features available with the add-on and how to use them.

Summary

The Advanced Facial Attribute Detection add-on powered by Cloudinary and the Face API of Microsoft's Cognitive Services provides a high precision mechanism that can analyze images and create the best crop for most sites as well as automatically add the nice artistic effects of exact overlay placing. The integration within Cloudinary's pipeline is seamless and dynamic using simple manipulation URLs.

We are excited to enter into this partnership with Microsoft's Cognitive Services so give the add-on a try. The Advanced Facial Attribute Detection add-on is available to all our free and paid plans.

Recent Blog Posts

Hipcamp Optimizes Images and Improves Page Load Times With Cloudinary

When creating a website that allows campers to discover great destinations, Hipcamp put a strong emphasis on featuring high-quality images that showcased the list of beautiful locations, regardless of whether users accessed the site on a desktop, tablet, or phone. Since 2015, Hipcamp has relied on Cloudinary’s image management solution to automate cropping and image optimization, enabling instant public delivery of photos, automatic tagging based on content recognition, and faster loading of webpages. In addition, Hipcamp was able to maintain the high standards it holds for the look and feel of its website.

Read more
New Image File Format: FUIF: Why Do We Need a New Image Format

In my last post, I introduced FUIF, a new, free, and universal image format I’ve created. In this post and other follow-up pieces, I will explain the why, what, and how of FUIF.

Even though JPEG is still the most widely-used image file format on the web, it has limitations, especially the subset of the format that has been implemented in browsers and that has, therefore, become the de facto standard. Because JPEG has a relatively verbose header, it cannot be used (at least not as is) for low-quality image placeholders (LQIP), for which you need a budget of a few hundred bytes. JPEG cannot encode alpha channels (transparency); it is restricted to 8 bits per channel; and its entropy coding is no longer state of the art. Also, JPEG is not fully “responsive by design.” There is no easy way to find a file’s truncation offsets and it is limited to a 1:8 downscale (the DC coefficients). If you want to use the same file for an 8K UHD display (7,680 pixels wide) and for a smart watch (320 pixels wide), 1:8 is not enough. And finally, JPEG does not work well with nonphotographic images and cannot do fully lossless compression.

Read more
 New Image File Format: FUIF:Lossy, Lossless, and Free

I've been working to create a new image format, which I'm calling FUIF, or Free Universal Image Format. That’s a rather pretentious name, I know. But I couldn’t call it the Free Lossy Image Format (FLIF) because that acronym is not available any more (see below) and FUIF can do lossless, too, so it wouldn’t be accurate either.

Read more
Optimizing Video Streaming and Delivery: Q&A with Doug Sillars

Doug Sillars, a digital nomad and a freelance mobile-performance expert, answers questions about video streaming and delivery, website optimization, and more.

Doug Sillars, a freelance mobile-performance expert and developer advocate, is a Google Developer Expert and the author of O’Reilly’s High Performance Android Apps. Given his extensive travels across the globe—from the UK to Siberia—with his wife, kids, and 11-year-old dog, Max, he has been referred to as a “digital nomad.” So far in 2018, Doug has spoken at more than 75 meetups and conferences!

Read more