Intuitivo, a pioneer in retail innovation, is using cloud-based AI and machine learning to revolutionize shopping. Their autonomous points of purchase (A-POPs), or vending machines, offer enhanced customer experiences at a lower cost compared to traditional vending machines. They use AWS Inferentia to run computer vision models faster and reduce costs by 95%. They also discuss the use of foundation models for product labeling and their ongoing partnership with AWS in driving retail transformation.
Revolutionizing Shopping with AI and Machine Learning
Intuitivo, a pioneer in retail innovation, has developed a cloud-based AI and machine learning (AI/ML) transactional processing system that is transforming the way customers shop. Our solution, called autonomous points of purchase (A-POPs), offers enhanced customer experiences at a fraction of the cost compared to traditional vending machines. With easy setup, maintenance-free operation, and a significantly lower cost, Intuitivo’s A-POPs are revolutionizing the retail landscape.
The Changing Retail Landscape and the Need for A-POPs
In the rapidly evolving retail industry, customers expect the same seamless shopping experiences they have when shopping online. Intuitivo believes that the future of retail lies in creating highly personalized, AI-powered A-POPs. These autonomous points of purchase bring products within arm’s reach of customers, providing a seamless shopping experience without long lines or complex transaction processing systems.
Cloud Inference for Real-time Product Identification
Intuitivo’s camera-based product recognition and payment system uploads transaction videos to the cloud for processing. Customers start a transaction by scanning the A-POP’s QR code, which unlocks the system. They can then grab what they want and go. Preprocessed videos are automatically processed in the cloud, and the customer’s account is charged accordingly.
Unlocking High-Performance and Cost-Effective Inference with AWS Inferentia
To optimize performance and cost, Intuitivo deploys its computer vision models on Amazon EC2 Inf1 instances powered by Inferentia. Inferentia is Amazon’s ML silicon designed for deep learning inference workloads, significantly reducing inference costs. By using Inferentia, Intuitivo’s computer vision models now run five times faster, achieving over 120 frames per second. This not only improves transaction speed but also enhances the accuracy of product detection within the models, delivering a superior shopping experience for customers.
Easy Data Parallel Inference with AWS Neuron SDK
Intuitivo maximizes the performance of Inferentia by using the torch.neuron.DataParallel() API from the Neuron SDK. This API allows them to fully utilize the available NeuronCores in the Inferentia accelerator, improving the performance of their inference workloads.
Accelerating Retail Transformation with Foundation Models and Scalable Deployment
Intuitivo recognizes the importance of foundation models in the retail industry. These models enable quick and accurate product labeling, enhancing the efficiency of A-POP solutions. Intuitivo has implemented the Segment Anything Model (SAM), a vision transformer foundation model that can segment any object in any image. SAM accelerates the labeling process and reduces training time and cost for product planogram models.
Conclusion
Intuitivo’s partnership with AWS Inferentia has resulted in a five times increase in processing speed and a 95 percent reduction in inference costs compared to their previous solution. This innovative technology is revolutionizing the retail industry, providing customers with seamless shopping experiences and retailers with cost-effective and scalable solutions. To learn more about how Inferentia can optimize your inference applications, visit the Amazon EC2 Inf1 instances and Amazon EC2 Inf2 instances product pages.
