4 Real-Life Cases of Innovations in Computer Vision Powered by AI

As the cloud model has gained momentum, the adoption of AI has parallelly increased and has become increasingly mainstream across a multitude of areas and applications.

In 2012, Google built a neural network simulating human brain learning that analyzed 10 million videos on YouTube to find cats. With 74.8% of accuracy when identifying cats, the neural network, as you may guess, found plenty.

The remarkable part of this research is that without giving any information about what a cat looks like, it could identify them in a video. “It basically invented the concept of a cat,” Google fellow Jeff Dean told the New York Times.

The latest research indicates that there are many computer vision applications in business, retail, healthcare, education, military, and more.

What Computer Vision is & How Does it work?

Computer Vision is a subset of Artificial Intelligence (AI) that uses deep learning. To emulate human sight, computer vision teaches machines to see things in images by breaking them down into pixels and identifying matches.

Each image needs to be tagged with metadata that indicates the correct answer. A neural network runs through the data and signals that it has found an image with a cat. The feedback received on whether the image was correct or not, helps it improve. Neural networks are using pattern recognition to distinguish many different pieces of an image. – Forbes

Examples of Computer Vision That is Driving Innovation

1. Predictive Maintenance in Aircraft

Aircraft maintenance is a serious business. Here, AI has the potential to transform the industry with its capabilities to expand, automate, and sharpen the collection and processing of intelligence. In fighter jets or even commercial aircraft, the higher-level model depends on the lower level. If those lower-level models are wrong, they can result in serious inaccuracies in a simulation process.

Predictive Maintenance (PMx) leverages advanced AI techniques to analyze both real-time and historical data. Using this technique, we can detect early symptoms of breakdown and prevent it. It improves aircraft readiness and reduces costs by minimizing the downtime for unscheduled maintenance.

Delta, the airline company, is investing in new tools and technology to further eliminate maintenance cancellations and enhance the customer experience. Between 2013 and 2017, Delta TechOps jumped from 169 cancel-free days to 324. Read more.

2. Detecting Malware with Computer Vision and Other Deep Learning Techniques

Researchers found that malicious files tend to often include ASCII characters of various categories, presenting a colorful image, while benign files have a clearer picture and distribution of values.

Visual representations of malicious (a), (b) and benign (c), (d) executable samples.
Visual representations of malicious (a), (b) and benign (c), (d) executable samples.

In 2019, researchers at the University of Plymouth and the University of Peloponnese published a paper demonstrating that when benign and malicious files were visualized by transforming binary and ASCII values into color codes, new patterns emerge that separate malicious and safe files. These differences would have gone unnoticed using classic malware detection methods. (Excerpt taken from this source)

3. The Future of Fashion is Data-Driven

Fashion is a fascinating domain for computer vision problems. Not only does it offer a challenging testbed for fundamental vision problems—human body parsing, cross-domain image matching, and recognition—but it also inspires new problems that can drive a research agenda, such as modeling visual compatibility, interactive fine-grained retrieval, or reading social cues from what people choose to wear. (Excerpt taken from this source)

Scope of the intelligent fashion (Source: https://dl.acm.org/doi/fullHtml/10.1145/3447239)
Scope of the intelligent fashion (Source: https://dl.acm.org/doi/fullHtml/10.1145/3447239)

Here are some of the LATEST fashion brands that are at the forefront of technological innovation.

i) Alibaba:

Alibaba’s FashionAI brings AI to the fashion world and offers customers personalized mix-and-match recommendations. The deployment of FashionAI in a concept store is the latest showcase of how New Retail technologies can breathe new life into traditional offline retailing.

“FashionAI embodies our thinking of what the future of fashion retail could look like and an exploration of using technologies to better understand and cater to consumers’ fashion needs,” said Zhuoran Zhuang, Vice President, Alibaba Group. “With the latest AI technologies, like machine learning and computing vision, FashionAI can now recommend items that match your style. That gives imagination to consumers and injects new ideas into fashion brands and retailers to rethink their business and sales models. Leveraging AI in fashion, therefore, offers many untapped opportunities for fashion retailers.” (Excerpt taken from the official press release)

ii) Walmart

Walmart’s latest acquisition of the virtual clothing try-on startup Zeekit, a real-time image processing technology in combination with AI, computer vision, and deep learning, has introduced state-of-the-art technology to Walmart customers to deliver an inclusive, immersive, and personalized digital experience that will better replicate physical shopping.

 

 

The retailer is rolling out Zeekit technology on its Walmart app and Walmart.com as a “Choose My My Model” try-on feature. Read more about it at Walmart.com.

4. Autonomous Military Weapons

Computer Vision and its subset, facial recognition have improved dramatically in only a few years. According to a test by the National Institute of Standards and Technology (NIST) in April 2020, the best face identification algorithm has an error rate of just 0.08% compared to 4.1% for the leading algorithm in 2014. Source 1 and Source 2.

Some of the credible media sources revealed that it was facial recognition that was used to track down and identify one of the most notorious men – Bin Laden.

In 2017, the US Department of Defense (DOD) worked on Project Maven, formally known as the Algorithmic Warfare Cross-Functional Team (AWCFT). Maven’s focus was/is to apply computer vision algorithms to tag objects identified in images or videos captured by surveillance aircraft or reconnaissance satellites. The program received national attention after Google Inc., one of several technology companies participating in the program, publicly withdrew amid uproar from employees about the “weaponization” of artificial intelligence (AI).  More on that here and here.

Conclusion

Computer Vision along with other cutting-edge technologies such as deep learning will continue to be extensively used. They will add more capabilities to the existing systems to deliver more intelligent and reliable outcomes.

There is no dearth of cutting-edge innovations coming in almost every day. So I am expecting more such innovations to be underway that can change how we live, work and communicate with each other.

 

By Bablu Kumar

Bablu Kumar is a technology writer with a focus on cybersecurity and the IT domain at large. The topics he writes about include AI and automation, malware, data breaches, exploits, and security defenses, as well as research and innovation in information security. Feel free to connect with him at https://www.linkedin.com/in/hacback17/

Leave a Reply

Your email address will not be published.