Cloudinary Blog

Face detection based cropping

by Nadav Soferman


Cloudinary’s roots go back to when we needed to embed users’ uploaded images and profile pictures for web projects we’ve developed.

We had to show people and faces of users in various dimensions and perspectives, and the process proved cumbersome.

These days, Cloudinary allows you to easily resize and crop your images to snugly fit the graphics design. But sometimes, just resizing pictures using ‘fill’ crop and ‘center’ gravity just doesn’t cut it. This is a common problem with user profile photos.

 

So we thought it is about time we add some more brainpower into the cropping process. Of-course we could just let the users manually crop their own profile pictures, but that just wasn’t fun enough :)

We decided to add face detection based cropping. Facebook use it internally to find faces of users and crop the profile pictures accordingly. Google does it as well. What about the rest of us?

Fortunately, we didn’t need to invent the wheel. Some top-notch face detection open source algorithms are freely available today. They are just not production ready or easy to integrate. The good news - we did the heavy lifting for you. We integrated existing tools and libraries, optimized their settings and added smart face detection based cropping to Cloudinary. Try it out - you won’t believe the difference that it makes.

Cloudinary supports either a single face detection or multiple faces detection (i.e., focusing on the people in a picture). You can easily create a face thumbnail of your users, crop an image to focus on a person or resize an image to fill the required dimensions while making sure the person in the original picture appears in the resized version of the image. And as usual, Cloudinary does all this automatically for you in the cloud and delivers the images through a fast CDN.

 

 

Using face detection based cropping with Cloudinary is very simple. Below are examples showing both cropping via our universal dynamic URLs and our Ruby on Rails integration.

http://res.cloudinary.com/demo/image/upload/w_100,h_100,c_thumb,g_face/face_top.jpg

<%= cl_image_tag("face_top.jpg", :width =>100, :height => 100,
:crop => :thumb, :gravity => :face) %>

For more details and usage examples, take a look at our documentation: http://cloudinary.com/documentation/image_transformations#face_detection

This feature is now available across all our plans, including the free one. We hope you like it.

 

Recent Blog Posts

How mindbodygreen reduce costs & save development time

mindbodygreen is a digital media brand dedicated to health and wellness. The company’s website bring together leaders in the wellness world, best-selling authors, fitness experts, trusted healers, pioneering doctors, top chefs, celebrities and top-tier journalists to provide readers fresh, expert content and tools to help them live healthy, happy lives. The company was founded in 2009 and is headquartered in Brooklyn, N.Y. 
Read more

Why JPEG is like a photocopier

by Jon Sneyers
Why JPEG is like a photocopier (generation loss)

If you make a copy of a copy of a copy, the quality will deteriorate with every ‘generation’. This problem is called ‘generation loss’. It is not difficult to understand why this happens with actual copier machines. Scanning and printing are not perfect, being based on noisy sensors and physical paper and ink, and the resulting noise will tend to accumulate.

Read more
Apply conditions to dynamically manipulate images.

It's great to have the capability to manipulate images on the fly by using dynamic URLs to customize the images to fit the graphic design of your site or mobile application. However, what if you want to manipulate an image depending on a specific image characteristic (like its width or aspect ratio) or its contents (does it contain a face?). What you need is a way to apply a transformation to an image only if a specific condition is met. Take for example a situation where you have allocated space on your page for a user uploaded image with a width and height of 200 pixels. Furthermore, if the image contains a face you would like to zoom in and focus on the face itself, otherwise you would like to fit the entire image into the available space:

Read more
Moviepilot focus on core business and reduce costs
 
Moviepilot Inc. is a media company that hosts a website and social publishing platform designed to bring together movie and TV fans, and help studios trigger, fuel and amplify positive conversations about entertainment topics. The site invites fans to create content, such as writing articles and producing videos, which are then curated on the site and distributed to more than 30 million fans worldwide. The site boasts more than 20 million unique visitors per month, with 87 percent of visitors accessing the site on mobile devices. The company is headquartered in Venice Beach, Calif., and has an office Berlin, Germany, where it was originally founded.
Read more
Powerful facial detection for image manipulation

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):

Read more