One of the hot trends in artificial intelligence (AI) revolves around the use of deep learning (DL) technologies for image and video classification. These AI-driven applications use computer vision to classify or categorize an image or video file on the basis on its visual content.
So, what is deep learning? In a few words, DL is a subset of machine learning (ML), and one of the key building blocks for AI solutions. It uses artificial neural networks as the underlying architecture for training algorithms, or models. These brain-like networks use multiple layers to progressively extract higher-level features from raw input.
In a simple example, data scientists talk about training models to identify cats in images. If the models are shown enough images of cats, they learn to recognize the features that make a cat a cat. They also learn how to distinguish cats from dogs, horses, people and whatever else.
Compelling use cases
The use cases for image and video classification are virtually unlimited. Across a wide range of industries, organizations can put these AI-driven capabilities to work to streamline business processes, automate tedious manual tasks and gain insights in real time.
For example, grocers can now use computer vision capabilities to inspect and grade the quality and freshness of meats, seafood, produce and bakery goods. A case in point: Walmart has developed an AI system that inspects fresh food for signs of defects and spoilage. This system is helping Walmart monitor the temperature and freshness of produce and perishable foods, improve visual inspection at distribution centers, and route perishable food to the nearest store.1
Similarly, retailers can use AI to make sense of the deluge of video steaming in from surveillance cameras that are now virtually everywhere in retail stores. This flood of data can be too much for security personnel to monitor effectively. Store operators can solve this surveillance challenge by using deep-learning inferencing capabilities to analyze video streams in real time and detect behaviors that are often associated with shoplifting. With these capabilities, the AI-driven system does a lot of the heavy lifting for the security personnel on the backend.
Examples like these could go on and on, because the use cases for AI-driven image and video classification span the gamut of industries, from healthcare and life sciences to agriculture and manufacturing.
A few case studies
To make this story more tangible, let’s look at a few examples of how Dell EMC customers are leveraging AI-driven image and video classification in the course of their operations.
- Agriculture – AeroFarms, a New Jersey-based vertical farming company, is using ML, data analytics, the Internet of things (IoT) and related technologies to bring new levels of precision and productivity to its indoor farming operations. For example, AI is used for automating image recognition and classification in order to adjust plant nutrients, light and other factors. As noted in a Dell Technologies case study, these technologies are helping AeroFarms in its efforts to continually optimize growing conditions.
- Healthcare – At the University of California, San Francisco and its Center for Digital Health Innovation, researchers are working to add AI to the process of diagnosing tears in knee cartilage, or the meniscus. In this initiative, explored in an Intel case study, the research team has set its sights on developing and training a DL model that can examine MRI results, identify those that show signs of torn knee cartilage and, eventually, objectively classify meniscus tears. The ultimate goal is to develop an accurate, data-driven grading system of meniscus lesions that can provide results to patients immediately after scanning.
- Ecology – Many mangrove forests have been cleared to make way for agricultural ponds, urban expansion and other human uses. These valuable resources are also threatened by natural or indirect events and processes, including sea-level fluctuations. To create a baseline for historical and future comparisons, a research team at Aberystwyth University in Wales leveraged satellite datasets from Japanese and U.S. space agencies. The team then wrote a software program and trained an algorithm to identify mangroves in satellite images. As a Dell EMC case study explains, this project will help researchers and governments around the world characterize mangroves and detect changes in their extent from established baselines — which is one of the keys to protecting an invaluable natural resource.
The fuel that makes it all run
The process of training DL models is both compute- and data-intensive. Deep learning applications require massive amounts of data, fast compute and equally fast storage, along with a lot of memory and high‑bandwidth networking. This is the fuel that propels the AI application forward.
To meet this need, Dell EMC offers a growing portfolio of solutions designed and optimized for DL. These offerings include the new Dell EMC Ready Solution for AI – Deep Learning with Intel. It provides an optimized solution stack that includes all the hardware, software and services organizations need to get an AI system up and running quickly. The name of the game here is quick time to value for an optimized AI solution.
“Getting started and using these technologies can be complex, so we’re focused on making it simpler for organizations of all sizes,” says Thierry Pellegrino, vice president of HPC at Dell EMC, in a recent news release. “We’ve engineered Dell EMC Ready Solutions as tested and validated configurations that help our customers more easily and quickly benefit from HPC and AI technologies to reach their ultimate goals.”2
With DL technologies, organizations can deploy image and video classification applications to streamline and accelerate processes that would otherwise be extremely burdensome, if not impossible, to carry out. And today, all the technologies are in place for the deployment of image and video classification applications across the enterprise.
To learn more
Advancing the Frontiers of AI
Dramatic advances in data analytics and high performance computing capabilities have created a foundation for the adoption of AI-driven applications in the enterprise. However, these enabling technologies are only part of the AI story. The other part is the rise of smarter algorithms that can glean insights from massive amounts of data. In this series of posts, we explore these building blocks for AI solutions in enterprise environments.
1 Supermarket News, “Walmart introduces Eden, its high-tech fresh-food initiative,” March 1, 2018.
2 Dell Technologies news release, “Dell Technologies Simplifies Customers’ Path to Innovation with AI and High Performance Computing,” June 17, 2019.