Postal service turns to computer vision AI and advanced computing to improve delivery – GCN
Postal service turns to computer vision AI and advanced computing to improve delivery
To help process 7.3 billion packages per year – 231 per second – the U.S. Postal Service is using artificial intelligence.
Announced this month, the USPS deployed advanced computer systems to 195 mail processing centers nationwide to apply seven computer vision models, reducing the time it takes to track a missing package from several days to less two hours. Additionally, a computer vision task that would have taken two weeks on servers with 800 processors can be completed in 20 minutes on four NVIDIA V100 Tensor Core GPUs in a Hewlett-Packard Enterprise Apollo 6500 server.
“Eight hundred servers is actually the equivalent of a data center in each of the 195 data centers. He’s absolutely a non-starter, ”said Anthony Robbins, vice president of North American Public Sector business at NVIDIA, in a virtual press conference announcing the work.
This is all based on the agency’s use of the Edge Computing Infrastructure Program (ECIP), a distributed AI system running on the NVIDIA EGX platform. Each edge server processes 20 terabytes of images per day from over 1,000 mailing machines. NVIDIA’s open source software, Triton Inference Server, provides the AI models that every mail center needs.
Thirteen EGX systems operate in two data centers and form the seven algorithms developed by USPS to date.
“The overall design here is to continue to improve and build a database for the packages so that over time they can improve the processing and efficiency of the packages and build from this model,” Robbins said. .
The project started about a year ago and USPS is already working on additional applications. For example, it launched a request for whoever uses optical character recognition (OCR) to streamline imaging workflow, according to a May 6 blog post.
“In the past we would have bought new hardware, software – a whole infrastructure for OCR; or if we were using a public cloud service, we had to upload images to the cloud, which takes a lot of bandwidth and costs a lot when you’re talking about a billion images, ”Todd Schimmel, the manager who oversees USPS systems, including ECIP, said in the blog.
The OCR use case will live as a deep learning model in a container on ECIP managed by Kubernetes and served by the NVIDIA Triton inference server, software capabilities that help organizations manage the deployment of the OCR. ‘AI on the periphery.
USPS entities have ideas for up to 30 applications for ECIP, according to the blog, and Schimmel hopes to launch some this year. One would automatically check if a package has the correct postage for its size, weight and destination, while another, which could go live this summer, would decipher damaged barcodes.
The USPS, long under siege for funding and operations, has several IT initiatives underway to address these concerns. For example, he is studying the use of autonomous vehicles for mail delivery and the monetization of his geolocation data. A report released in February by the Information Technology and Innovation Foundation touted the use of robotics for last mile postal delivery.
While NVIDIA’s current work relates to parcel processing only, the framework is in place to develop capabilities for the full range of mail processing, Robbins said.
“Not only is extraordinary innovation happening within the US Postal Service, but the work done in creating a company-wide AI program for the US Postal Service may actually be a motivation for the US Federal Government. and, frankly, for business ventures and businesses around the world. world, ”he added.