Originally posted at Spiceworks News & Insights
The massive improvement in computing capabilities over decades means AI can contribute to numerous fields requiring high computing capabilities. Here, Ned Hill, founder and CEO, Position Imaging, discusses how AI and computer vision can significantly contribute to our everyday lives.
There was a time when the idea of robots capable of thought was only a Hollywood plot device used in blockbuster films like 2001: A Space Odyssey, Terminator, Blade Runner, and The Matrix to keep us engaged. But true Artificial Intelligence or AI is now becoming fact over fiction, much like Hollywood movies depicting humanity walking on the moon in the 1950s. However, computing capabilities need to mature first.
Past and Present Capabilities
The first software to approximate human-like skills and problem-solving was under development in the mid-1950s. As the concept of AI evolved over twenty years, the problem was not programming but rather a lack of computer storage holding it back. To develop, AI needed higher processing speeds.
Using the increasing processing power and data storage, AI engineers began to further the field of research to include supercomputers such as Indiana University’s Big Red 200, which has 672 compute nodes. Each has 256 GB of memory and two AMD EPYC 7742 processors with 2.25 GHz and 225 watts.
Alternatively, the HiPerGator 3.0 from the University of Florida has 240 AMD EPYC Rome workstations with a RAM of 1024GB and more than 30,000 cores and 150 AMD EPYC Milan machines with 512GB of RAM and more than 19,000 cores in total. So now is the ideal time to make AI mainstream.
Applications requiring more computing power than most devices can manage today are great candidates for AI systems. For instance, in healthcare, machine learning can group similar medical cases to ascertain patterns, leverage current data to make a prognosis or even manage appointment scheduling and other data management tasks.
See More: How To Reduce Bias in Machine Learning
According to a recent Forbes article, the insurance industry is leveraging machine learning to help detect fraud, “In the analysis of potential fraud, multiple tools are used, some are in ML [Machine Learning], such as statistics, rules-based approaches and even neural networks.” And Towards Data Science says that many financial companies are already using machine learning to “reduce operational costs, increase revenues and reinforce security.”
Even the environment will benefit from the intellectual output from supercomputers. For example, IDC points out, “Thanks to new research on a supercomputer, Ramgen will begin testing a 13,000-horsepower CO2 compressor this year. This compressor is projected to reduce the capital costs of CO2 compression by 50% and produce a minimum of 25% savings in operating costs. Applying these cost savings to a new 400-megawatt clean coal plant would result in capital cost savings of approximately $22 million and an annual operating cost savings of approximately $5 million.”
AI Gets the Hype, but Computer Vision Has Insights
While AI gets all the hype, businesses are now using thousands of computer vision applications to expedite or automate numerous processes in fields like healthcare, where computer vision can spot cancer from CT images faster than doctors can.
Computer vision is a branch of AI employing data to detect and recognize objects seen through a computer’s camera. This cousin of AI recognizes objects and their environmental conditions through a camera lens and gives the computer a digital grasp of its surroundings to interact with objects. For example:
- Individuals can be uniquely identified in highly secure situations using fingerprint and retinal scanning to grant or restrict access.
- Problems in wind turbines can now be seen in advance using autonomous drone footage with high-definition cameras.
- Computer vision applications help track packages or products through the supply chain.
Computer Vision’s Practicality in Everyday Living
We often notice the practical computer vision applications first regarding life efficiencies. Some organizations, for example, use computer vision technology to assist multi-family property managers in automating the package handling process. In addition, they use it to redirect workers to manage residents rather than deliver or sort items received from couriers.
Residents will benefit from the application in this scenario because they will no longer have to wait for personnel to pick up parcels. In addition, couriers benefit from multi-family computer vision applications since they can deliver packages directly to a smart package room. The computer vision technology in this intelligent package room virtually monitors and tags the location of each box.
Besides multi-family housing, logistics companies employ computer vision to audit packages passing through their hubs. This computer vision allows senders to assess package dimensions correctly before delivering them. Logistics firms may improve the customer experience and lower costs by automating the measurement of packages. Py-tesseract is a tool that allows for data extraction from images.
These computer vision efficiencies become increasingly crucial as AI applications migrate to operate drones, mobile devices, and vehicles. Qualcomm’s AI accelerator architecture in the Hexagon 780 Processor is furthering AI research. In addition, its Gauge Equivariant Convolutional Neural Networks are improving the detection skills of computer vision systems.
Qualcomm’s neural networks increase curved shape detection, allowing computer vision to recognize item dimensions better. This work will undoubtedly contribute to the adoption of AI in daily gadgets and IoT networks and the performance of computer vision applications.
Unlike the drama depicted in most Hollywood films, people will continue to unknowingly benefit from AI and its computer vision relative to performing commonplace tasks such as presenting the next Netflix movie to watch or a smart package room notifying the correct person to receive a recently delivered Amazon package. Intelligent computing applications also go beyond superficial conveniences to perform complex duties such as driving, weather predictions, and rebalancing our retirement portfolios. And the best is yet to come.
The trickle-down impact of these billion-dollar academic supercomputers will be realized through improved national security, redesigned components that promote engine fuel efficiencies and reducing heart disease that will prolong our life expectancy. The application of AI and computer vision is both practical and surreal — already dramatically impacting every aspect of our lives, it will continue to reveal itself in subtle and surprising ways for a long time to come.