The Power of Image Search for Talent Sourcing was originally published in the excellent book ‘Social Media Recruitment. How to successfully integrate social media into recruitment strategy’ by Andy Headworth (2015). Although most of the article’s content still applies for image search to identify talent today, not all of the predicted developments have taken off as I would have hoped, mainly as a result of privacy regulations. In the next couple of weeks I will rediscover Image Search for Recruitment, and see what has changed since I wrote the Power of Image Search, almost a year ago. To get a copy of Andy’s book and learn more about how social media can be successfully integrated into recruitment, visit this link. Below the original article.
Images tell the story
In our increasingly visual world images tell the story. Social networks have been early in recognizing this, together with the human urge to post, share, view, like and comment on photos. Images are the number one driver behind social network growth, helped by features that will make users upload even more photos, such as (auto) tagging and ‘awesomizing’ features.
In 2014 over 1.8BN photos were uploaded and shared PER DAY on platforms such as Whatsapp, Facebook, Instagram, Snapchat and Flickr. In late 2013 Yahoo already predicted the number of photos taken in 2014 to approach a staggering 1 trillion thanks to the selfie explosion. We are talking here about ‘big data’ sets.
To deal with this amount of pictures on the Facebook network, for example, Facebook alone has built three ‘cold storage’ data centers to store less popular or outdated photos. Each of their 16,000 square-foot data centers is able to hold an exabyte of data, similar to about one million computer hard drives. These images will not only be stored and indexed to easily retrieve them. From their computer vision systems back-end – machine learning systems that have the power and intelligence to identify what is in an image, what a building looks like versus a face versus a landscape – social networks will use images to gain intelligence about what we are doing, who we are hanging out with and what our interests are. This data will most likely be of interest for their advertisers.
And that’s not all. More cameras, mobile phones, apps and social networks use GPS technology to exactly determine where photos have been taken. This information is stored in Exif (Exchangeable Image File Format) files, small data files embedded in images. Companies such as Facebook, Instagram, Google and Foursquare get access to this data as soon as their users post images to their networks. Have you ever wondered why social networks would like you to tag who is in your pictures?
A talent sourcer’s goldmine
With all the information available in images, how valuable are images for talent sourcers? After all, sourcing is about finding people and gathering information about people. With the amount of photos that people have all over the internet, sometimes without even knowing of their existence, sourcers have access to an invaluable source of information. Images have become a sourcer’s goldmine.
As a result of the popularity of using images, image search indeed can be a very powerful instrument for talent sourcing. A great starting point of image search is using avatar pictures or profile images. These images usually contain a person’s face and since most people have a habit of using one single image for different online profiles, it makes it relatively easy to find all the social networks a person is active in by simply conducting a search on the person’s profile image. Interestingly, one of the benefits of using image search is that it can deliver more relevant results as opposed to just name searching. Especially for more common names, images prove to better in identifying a unique person.
One of the best ways to conduct an image search is using the reverse image search technique in Google Images. Most of the other examples of image search are just variations on reverse image search.
There are basically four ways to search by image using Google Images:
- Drag and drop: drag and drop an image from the web or your computer into the search box on http://images.google.com.
- Upload an image: on http//images.google.com, click the camera icon, then select ‘Upload an image’. Select the image you want to use to start your search.
- Copy and paste the image URL: right-click an image on the web to copy the URL. On http://images.google.com, click the camera icon, and ‘Paste image URL’.
- Right-click an image on the web: to search by image even faster, download the Chrome extension (https://chrome.google.com/webstore/detail/dajedkncpodkggklbegccjpmnglmnflm?hl=en) or the Firefox extension (https://addons.mozilla.org/en-US/firefox/addon/search-by-image-by-google/). With the extension installed, simply right-click any image on the web directly and select “Search Google with this image” to initiate the image search.
Both Chrome and Firefox browsers have multiple reverse image search plugins and extensions available. To install these visit their web stores and search for ‘image search’ in the search bar. Most of these plugins are based on Google’s image search technology. It can be worth trying different tools in different browsers though, as search results may vary.
Applying either of the above techniques to an image containing a person’s face will result in Google finding similar pictures of that person and redirecting you to the profiles containing the pictures. An additional step to take in this search is to click on the Find other sizes of this image link(s) right next to the initial search result, to find more results with the same image, only in different sizes.
Social networks all have set their specific image dimensions for the different parts of the social network where profile images are being used and all image sizes are stored separately, making it more convenient to find multiple online profiles of a person by using a single image.
Once a search engine is able not only to identify similar images, but also is able to establish the name of the person who the image is from, search results get even more interesting and often show other images of the same person. The results then usually also contain images of other social profiles connected to the same person, which subsequently can be used to further explore by simply right-clicking the image using a reverse image search plugin again.
The same technique can be applied not only to profile images, but to any other image that contains a face. Think of images on company introduction pages (meet the team), event pictures, pictures used for online check-ins on Foursquare, photo sharing communities such as Flickr, blogger profiles and much more.
Uploading an existing image with a person’s face to one of the search engines is just one method. Alternatively, interesting results can be obtained by cropping a person’s face from a larger image of a group of people and uploading the cropped image to the search engine.
Google is not the only search engine offering image search technology. Some other search engines that have image search functionality enabled and are worth exploring for image search purposes are:
- http://yandex.com/images/ (Russia)
- http://image.baidu.com/ (China)
Other search engines are dedicated to reverse image search, some of which offering more advanced image search utilities:
The future of visual sourcing
In the near future image search technology will become far more advanced, as social networks and search engines will benefit from facial recognition technology becoming more accurate. Most of this technology is already available and used by governments to prevent crime, but commercial use still is very much restricted by privacy regulations. The major breakthrough of image search for talent sourcing will most likely be wearable devices having facial recognition technology enabled and linking this to all image data sets and social networks to allow visual sourcing on-the-go.
(TruLondon Crew photo credit: Sara Headworth)