Picaas (Picture as a service) uses AI to remove image watermarks
iKala say they’re a Human-Centered AI marketing technology company, but what’s important is their tech. They’ve announced what they claim is the world’s first smart image optimization solution for Google Shopping Ads “Picaas” (Picture as a service).
To sell products on Google Shopping and then promote those products with ads, you have to meet certain image requirements. In this video by Google for Retail, Google strongly recommends that you use product images that are clean, clear and shown a single item in each image, with a simple background, the highest resolution possible.
Imagine you don’t have clean images of all your products, but you do have access to some from your web team. They may include watermarks, promotional sale overlays etc. What Picass does is to analyse the photo’s contents and based on the brain established by massive collections of similar images, can fill in the contents behind the object or text removed. This works like Adobe’s content-aware fill in Photoshop, just on the web and really fast.
Picaas enables companies running Google shopping campaigns to do mass product feed image adjustment, eliminating non-compliant elements and enhancing product image in just 2.2 seconds. The solution is now available globally and offers a time-limited, free tryout campaign.
Consulting firm Adthena reports retailers across the United States spend 76.4% of their advertising budget on Google Shopping Ads.
“Google is aggressively promoting Google Shopping Ads around the globe and providing e-commerce companies with a conversion rate that’s two to three times higher than ordinary digital ads.
However, we’ve seen many companies spending extra time tackling with the strict product feed requirements. Picaas is born under the concept to automatically solve those issues and help companies to build a successful Google Shopping campaign faster, cheaper and more effective.”
iKala CEO Sega Cheng
You can try the technology for free at https://picaas.io/demo.html, just press escape to see the results. I wanted to push the tech and see how it went with less of a product image and more of a regular web graphic with text overlay.
You can see it does a pretty good job, but did have an error on the man’s face, much of which was obscured by the text. After trying again on another image of sneakers with text outside the product, it worked flawlessly.
The Picaas AI automatic image solution combines multiple deep-learning models. The basic principle is to use image segmentation technology to detect portions of the ad that do not meet the requirements. The software then uses image inpainting technology to complete the picture and improve its overall quality, so the rejected product picture can be successfully approved and delivered to potential customers.
“Picaas exemplifies the mission of artificial intelligence, which is to help humans take care of time and effort consuming tasks. Picaas was able to achieve a success rate of 70% in the early stages of its development, so its future potential of commercializing the field of AI-based computer graphics is to be anticipated.”
Professor Bing-Yu Chen, Associate Director of the National Taiwan University IoX Center
Picaas utilizes Google Cloud Platform (GCP) to effectively process a massive number of product pictures simultaneously. This satisfies the need for e-commerce platforms to deliver product advertisements in bulk with one click of the mouse.
The traditional method of manually editing Google Shopping Ads images took an average of fifteen minutes to process one picture. With Picaas, in best-case scenarios, a picture can be automatically edited in 2.2 seconds, which is four hundred times faster than humans. Currently, the minimum price per picture is around 0.1 US dollars, which is 50% to 95% cheaper than the traditional method of using human labor to edit pictures.
Picaas are running a promotional campaign now that offers automatic optimization of three hundred Google Shopping Ads product images for free, for the first twenty participants signed up on the Picaas official website from June 27th to July 11th of 2019.
For more information and event details, please visit: https://picaas.io/