5 Amazing Use Cases for Image Analytics
06/13/2018 by Ivan Gomez Modernization - Analytics
The applications of image analytics are endless. Companies are starting to realize the possibilities of how to extract value from unstructured data. Using images or videos, they can create a new and enticing customer experience within the retail, entertainment, insurance claims, and more.
Here are five image analytics applications that are unexpected, disruptive, and creative.
Curious to know who attended the Royal Wedding? Sky News partnered with Amazon.com and engineering firms to identify the attendees of Prince Harry and Meghan Markle’s wedding. They identified celebrity guests using real-time artificial intelligence (AI) capabilities. With Amazon Recognition, they compared live video footage of the guests entering the chapel against known, archived celebrity facial images.
In addition to identifying celebrities at a Royal Wedding, image recognition is being used to support an ever-growing list of business use cases. These practical applications of deep learning and image analytics are due to advances in machine learning algorithms, dataset availability, and the existence of robust technology with platforms to support real-time processing. Because of these advances, image analytics is now a realistic possibility for a growing number of organizations.
According to USA Today, TSA is investing in new scanners that allow agents to “virtually unpack bags.” These scanners would provide more accurate object detection, reduce the number of bags that would need to be opened and inspected, and provide faster security screenings.
Many US airports are acquiring upgraded technology that enables the use of biometrics such as finger or iris scanning, as an alternative security screening measure. Singapore’s Changi Airport will soon be opening a new terminal with automated face recognition. Since the Changi Airport is considered to be “the 6th busiest airport for international traffic,” they must be efficient.
By using image analytics technology, they expect to improve the airport’s ability to move passengers from arrival through security to their departure gate and significantly boost capacity.
Social media platforms, such as Facebook and Google Photos, have used deep learning and facial recognition for a while now. Whether you realize it not, all of us who use these platforms are helping them improve the accuracy of their models. If you’ve ever tagged a friend or family member, you’ve contributed to refining the model’s ability to detect individuals in the photos you post.
Facial recognition technology is also being used in Australia to identify missing persons. The Missing Persons Action Network (MPAN) is leveraging Facebook as a quick way to spread a message through be-friending missing persons to expand their network through Invisible Friends. With Facebook’s facial recognition algorithms, they can identify people in the background of photos. Because of how interconnected friends of friends’ networks are, it becomes possible to find missing persons. That is a social media analytics best use case!
Have you ever gotten into a car accident and had to go through a week-long claims process? Wouldn’t it be much easier – and less stressful – to be able to pull out your phone, take a few pictures of the damage, and upload them to an app for a real-time assessment?
Some insurance companies are already using AI technology. Mitchell Insurance has an automated vehicle damage analysis, allowing for more consistent and timely cost estimates. The company was able to assess totaled cars, heavy-damage vehicles, and light damage requiring only a paintless dent repair tool. This image data leads to better customer experience, reduced opportunities for fraud, and improved operations for the company. [Here are the technical details of how to use SAS Viya and Python to complete this image recognition task.]
There are countless examples of how to apply deep learning algorithms to healthcare. Drug manufacturing companies continuously design and test drugs, treatments, and devices through clinical trials. Clinical research can use deep learning with medical imaging (CAT scans, X-RAYs, and MRIs) to detect the presence or absence of conditions that are visible in images.
The CheXNet algorithm developed by a Stanford machine learning group can now detect pneumonia from chest x-rays with accuracy exceeding practicing radiologists.
This example shows how deep learning and AI are instrumental in improving health care practices and preventing false diagnoses. This technology can also provide physicians with a second opinion from a model trained on hundreds of thousands of images.