According to the Businessinsider, Users on Facebook are now able to search for images using key words that describe a photo’s contents, instead of relying on manual tags and captions, according to TechCrunch.
USING ARTIFICIAL INTELLIGENCE AND DEPP NEURAL NETWORKS
Using artificial intelligence and deep neural networks, the social giant is able to surface more contextually relevant photos than it could previously.
FACEBOOK NOW HAS BETTER UNDERSTANDING OF IMAGES AT THE PIXEL LEVEL
Facebook now has better understanding of images at the pixel level, and can recognize objects in an image, the type of scene it is, or if the image contains a well-known landmark, among others. While early, the search function will also help surface contextually relevant videos and other immersive formats. Here a few implications of the launch:
- It could enhance Marketplace monetization. Marketplace is Facebook’s peer-to-peer selling tool, and the object recognition could direct users to purchase items they see in a photo or video. For example, a search for “black shirt photo,” and will return pictures of that user (or their friends) in a black shirt, but Facebook could also show relevant results in Marketplace as well.
- The launch opens up the door for more search ad dollars. As of Q2 2016, users were conducting over 2 billion searches per day on Facebook. The company can monetize these by rolling out a paid search ad tool to monetize high-value intent-based queries. Paid search could drive more adoption od Dynamic Product Ads, which are currently targeted at users based on interest and browsing habits, but not a manual search.
- Competitors are rolling out similar smart search features. Pinterest is rolling out keyword search ads to a wider set of advertising partners, according to The Wall Street Journal. The company also recently launched “Visual Search,” allowing users to search for products within a Pin’s image. Meanwhile, Google’s open-source “TensorFlow” can identify and label images with an over 90% accuracy