In this special guest feature, Tom Setzer, Software Product Director at Accusoft, discusses how artificial intelligence will impact web developers, specifically relating to document management and imaging on the web. Tom joined Accusoft in 2006 as a project manager and is now the director of the SDK product group. His analysis of customer requirements and business data assists in the development of product strategies for the group. His 25+ years of experience includes 12 years of developing technology solutions for industry leaders including Motorola, Capital One and General Electric. Tom earned a BS in Computer Science from the University of North Carolina at Wilmington and an MBA from the Fuqua School of Business at Duke University.
There have been a lot of advancements in recent years in the field of artificial intelligence (AI), with real world, day-to-day impacts on our lives. Speech recognition in our hand-held devices, self-driving cars, robotic vacuums, and movie recommendations from Netflix are just a few examples. While these advancements have obvious impacts on consumers’ daily lives, they also impact web developers in a number of ways.
One example that is starting to impact developers comes from Google. The tech giant recently announced a new imaging technique to scale up an image using AI, known as upsampling. While the concept of using AI for image scaling isn’t new, Google’s solution is much faster due to some interesting tricks under the hood. Their solution, called RAISR (Rapid and Accurate Image Super Resolution), could allow Google to upload and store smaller versions of your image and bring them back to normal size on-demand, saving significant amounts of storage and bandwidth costs.
With “nearest-neighbor interpolation” and other simplistic scaling approaches that don’t leverage AI, the resulting image is larger than the original and preserves all the original detail, but typically has jaggedness.
Google’s super resolution is performed using learned filters to fill in details when scaling, instead of using interpolation, resulting in a more accurate and better-looking image. These filters are created by training processes against thousands of low-resolution input images and their high-resolution ground-truth counterparts.
In addition to upsampling images, web developers have other AI solutions in their toolboxes. For e-commerce sites, or other sites that want to make recommendations, there are commercial and open-source recommendation systems that use machine-learning. These systems learn what a user likes through their behavior, and then subsequently, make recommendations.
In the realm of image processing, AI has been used for years in a wide number of ways, and we’ll continue to see developers harness this technology in the coming years. Optical character recognition (OCR), handwriting recognition and handprint recognition often use some form of neural networks to convert images into text. Object recognition is used by Google, and others, to identify objects, such as a dog or a hat (or even a dog wearing a hat) in an image. This allows Google to present the correct images when users are searching the web.
And of course, the biggest impact to web developers is Google’s algorithm to rank pages and return the appropriate web pages based on a user’s search. Google continues to advance algorithms from simplistic metadata and link information to utilize information learned about the page and actual content.
Web development is a constantly changing role, and there are always a multitude of new techniques and disruptions that keep developers on their toes. Artificial intelligence is just the latest breakthrough shaking up the role of developers. As AI continues to advance for web development operations, developers will need to roll with the punches and leverage its abilities.