How Computer Vision and AI are Transforming Retail Technology

Many retailers operating brick-and-mortar stores are focused on creating convenient, frictionless shopping experiences while also improving their operational efficiency.  

With competition from online shopping and ongoing challenges such as loss prevention, stores are finding that advancing computer vision, AI, and IoT solutions can help address these issues. Additionally, the technology can create real-time promotions that are personalized to create more loyal customers. 

At Intel, we’re working with partners to offer optimized software, innovative hardware, and market-ready systems that help retailers differentiate themselves in ways that customers notice while also automating inventory and security management. 

Enhancing retail with computer vision and AI 

Let’s start at the end of most visits to a store: at the checkout line. Many stores have fewer cashiers, and the average self-checkout kiosk can create a hassle for shoppers. Long checkout lines can result in lost sales when shoppers decide a transaction isn’t worth the wait.  

A combination of computer vision, object recognition, and AI can solve that by empowering self-checkout kiosks that allow customers to simply set their objects on a counter. This already is a reality in places like Empower Field at Mile High. The stadium in Denver, Colorado, integrated AI-driven checkout kiosks into some of its concession areas. By using computer vision to generate 3D images of the products customers want to buy and matching them to an inventory database, the system can generate a total price in less than a second. The median transaction time is under 15 seconds. The stadium has seen a 34 percent increase in concession sales as well.  

Convenience stores and other retailers are adopting this type of AI-powered self-checkout to free up staff for other customer services. Alimentation Couche-Tard Inc. is deploying more than 10,000 Mashgin AI-powered Smart Checkout systems to over 7,000 of its Circle K and Couche-Tard convenience stores. As with the Denver stadium checkout system, customers can simply put their items down and pay as they normally would, speeding up checkout by up to eight times compared to traditional self-checkout. 

These types of systems can help track inventory without complication as well, which is a more common use of computer vision and AI in retail spaces to keep high-demand goods in stock, track sales trends, and help reduce theft and other loss of products. Smart camera systems can keep track of the number of sweaters left on a shelf and read barcodes on product boxes from a long distance.  

In addition to fixed computer vision systems, mobile robots also are starting to perform inventory tasks in stores as well. In only 30 minutes, one such Intel-powered robot can audit thousands of items in a store and note out-of-stock or misplaced items as well as incorrect or outdated sales prices. 

Using data collected from these systems, retailers can use analytics solutions to analyze in real time and over the long term what products consumers are buying quickly, and which products are languishing. Additionally, computer vision systems and AI can monitor for theft and alert employees or authorities as needed. Collecting and analyzing collected data can help retailers determine which high-value items are most likely to be taken and need to be secured. 

Data management and analytics systems, with assistance from AI, also can analyze a customer’s past purchases and reactions to past marketing campaigns to better tailor personalized promotions. Retailers can use this, for example, to give loyal shoppers special deals or create special marketing messages for fans of specific brands. 

How Intel and partners are helping retailers 

Our ecosystem of solution partners offers these and other innovative solutions for the retail sector using Intel technology that spans from the edge to the cloud. To power computer vision applications, Intel provides both hardware and software. Intel® Distribution of OpenVINO™ toolkit enables the development of vision applications on Intel platforms, including VPUs and CPUs, while Intel® RealSense™ technology is a reliable, multi-camera, vision-based system designed to understand the retail environment in 3D. This technology already is in use by mobile robots that can track inventory or automate order fulfillment.  

Our processors also power many retail digital solutions, as Intel® Xeon® Scalable processors can power high-performance analytics and include built-in features to accelerate AI. For retail solutions at the edge—including point-of-sale systems, interactive kiosks, digital signage, and more—Intel® Core™ and Intel Atom® processors come in a range of options. 

Intel® Arc™ graphics can power multi-monitor video walls in retail spaces and allow for the creation, sharing, and streaming of high-quality, AI-augmented graphics and images. The Intel vPro® platform also delivers performance, remote management, and security features to help ensure that critical retail devices keep running reliably.  

Learn more about the future of retail 

As retailers adjust operations to a more competitive, changed landscape for their stores, they’re trying to serve shoppers who want frictionless interactions and unique customer service experiences. Computer vision and AI tools can help deliver those things while also improving efficiency, reducing loss, and managing inventory. 

Intel can help with computer vision, AI, compute, storage, and networking technologies that let retailers collect, manage, and analyze data. We also work with a large ecosystem of partners that bring together our technology with their own market-ready innovations. Learn more about how we’re helping to transform retail operations.  

Article co-authored with @GeorgeLoranger

Publié dans Non classé | Commentaires fermés sur How Computer Vision and AI are Transforming Retail Technology

Food Sales Prediction Model using scikit-learn* (sklearn): Developer Spotlight

Developer Spotlight: Sayan Malakar proposed an AI solution for food sales prediction

Publié dans Non classé | Commentaires fermés sur Food Sales Prediction Model using scikit-learn* (sklearn): Developer Spotlight

Accelerating Codegen training and inference on Habana Gaudi2

Optimum Habana  makes it easy to achieve fast training and inference of large language models (LLMs) on Habana Gaudi2 accelerators. In this blog, we will walk through the process of performing Low-Rank Adaptation (LoRA) training of Codegen , an open-source LLM for program synthesis. We will also benchmark the training and inference efficiency of Habana Gaudi2 using Codegen

Publié dans Non classé | Commentaires fermés sur Accelerating Codegen training and inference on Habana Gaudi2

FAENet: Intel Labs and Mila Collaborate on Data-Centric AI Model for Materials Property Modeling

Intel and Mila collaborated on FAENet, a new data-centric model paradigm that improves both modeling and compute efficiency across different types of materials modeling datasets.

Publié dans Non classé | Commentaires fermés sur FAENet: Intel Labs and Mila Collaborate on Data-Centric AI Model for Materials Property Modeling

New Possibilities for Generative AI SaaS: Intel® Liftoff Startups at Intel® Innovation 2023

Intel Innovation 2023 at the San Jose Convention Center is a global tech gathering focusing on AI, future-proof platforms, and startups, including six promising ones from the Intel Liftoff for Startups program. It’s a must-attend event for tech enthusiasts and industry leaders.

Publié dans Non classé | Commentaires fermés sur New Possibilities for Generative AI SaaS: Intel® Liftoff Startups at Intel® Innovation 2023

Terrain Analytics Uses Intel® 4th Gen Xeon CPUs To Achieve Higher Throughput with Lower Costs

Terrain Analytics, a finalist in Intel Liftoff’s AI startup Hackathon, offers HR leaders valuable insights into talent management. Their platform enhances hiring precision and cost-effectiveness for businesses.

Publié dans Non classé | Commentaires fermés sur Terrain Analytics Uses Intel® 4th Gen Xeon CPUs To Achieve Higher Throughput with Lower Costs

Single Cell RNA Analysis using Intel® AI Analytics Toolkit: Developer Spotlight

Krittika Kanjilal proposed solution for single-cell RNA analysis using Intel® AI Analytics Toolkit

Publié dans Non classé | Commentaires fermés sur Single Cell RNA Analysis using Intel® AI Analytics Toolkit: Developer Spotlight

Startups pioneer LLM fine-tuning on Intel GPUs

Intel® Liftoff members showcase applications powered by the largest LLMs yet fine-tuned on Intel GPUs running on Intel’s new Developer Cloud.

Publié dans Non classé | Commentaires fermés sur Startups pioneer LLM fine-tuning on Intel GPUs

MatSci-NLP: Intel Labs and Mila Collaborate on Benchmark to Assess Materials Science Language Models

Intel and Mila collaborate on MatSci-NLP, the first broad benchmark for assessing the capabilities of language models on understanding materials science language and performing useful tasks for materials scientists.

Publié dans Non classé | Commentaires fermés sur MatSci-NLP: Intel Labs and Mila Collaborate on Benchmark to Assess Materials Science Language Models

Intel Labs Selected by DARPA H6 to Develop Tactical-Grade Clock with Microsecond Timing Precision

Intel Labs and collaborators from the University of Pennsylvania, Carnegie Mellon University, and IS4S have been selected by DARPA to perform in the H6 program to develop a GPS-independent tactical-grade clock with microsecond timing precision.

Publié dans Non classé | Commentaires fermés sur Intel Labs Selected by DARPA H6 to Develop Tactical-Grade Clock with Microsecond Timing Precision