In an increasingly interactive entertainment industry, media is rapidly shifting from text to video, making it more difficult to measure return on investment for advertisers and content buyers.
While global sponsorship spending topped $69 billion (€60 billion) in 2016, most sponsors only spend 1-2 percent of their sponsorship budget to measure ROI. Sports sponsors account for 70 percent of this.
According to Gartner, chief marketing officers will spend more on IT than chief information officers by 2017. CMOs desperately need real-time and responsive video analytics solutions to bring efficiency and transparency to sponsorship budget management.
The SAP Innovation Center Network (SAP ICN), which has a primary focus on machine learning and deep learning, is a big fan of NVIDIA. A recent announcement highlighted our relationship. Back in 2012, SAP R&D experimented with NVIDIA’s Kepler architecture cards to accelerate its HANA predictive analytics library. In 2014, we started our first deep learning experiments in the SAP ICN in Israel. Over the next year, we continued to fine-tune computer vision techniques and grow our team of experts.
In 2016, SAP invested in a new wave of machine learning applications. From experimenting with traditional tools and architectures like Caffe, TensorFlow, OpenCV and GoogleNet to CNN and RCNN, our team learned the transformational performance that AI and deep learning techniques brought to the table.
When we tested on NVIDIA GPUs, we knew immediately that would be our next platform for training and inference. Without hesitation, we placed an order for the first NVIDIA DGX-1 AI systems, which arrived in our labs in Israel and Germany last September. The normal product production cycle usually takes close to two years. Thanks to DGX-1, we were able to take our product to market within nine months! This would simply not have been possible without the availability of powerful GPUs.
The first solution we delivered was SAP Brand Impact, which automatically analyzes brand exposure in videos by leveraging advanced computer vision techniques and proprietary algorithms built by the SAP ICN computer vision team in Israel. It helps media agencies, production companies and brands gain accurate, timely insights into sponsorship and advertising ROI.
SAP Brand Impact deep learning models are trained on NVIDIA DGX-1 and the inference engine uses the TensorRT library. NVIDIA’s deep learning platform helped streamline SAP Brand Impact development and improve the runtime scalability.
Customers are excited about SAP Brand Impact. As a long-term SAP customer, Audi got early access to the latest SAP solution powered by NVIDIA deep learning, explains Global Head of Audi Sports Marketing Thomas Glas.
“Audi’s sponsorship team found the SAP Brand Impact solution a very useful tool. It can help Audi to evaluate its sponsorship exposure at high levels of operational excellence and transparency,” Glas said. “We were impressed by the capabilities and results of the first proof-of-concepts based on video footage from Audi FIS Alpine Ski World Cup. We’re strongly considering possibilities to combine SAP Brand Impact with our media analysis workflow for future Audi-sponsored events.”
As a club who’s shown excellence on the court for more than 40 years, winning six European championships, it has always been important to Maccabi Tel Aviv Basketball Club to be at the top level in Europe. It’s an effort that doesn’t stop off the court.
“SAP Brand Impact, powered by NVIDIA deep learning, provided us the opportunity to get access to the video analytics of the exposure our partners are getting in our games,” said Yaron Talpaz, the club’s chief marketing officer. “We are impressed by the solution’s capabilities, precision and speed and see it as a great potential tool for our sales efforts, adding data and numbers to media exposure previously deemed unmeasurable.”
To learn more about the solution, join us for a live webinar on Tuesday, Sept. 5, from 9-10am PT. Jim McHugh, vice president and general manager of NVIDIA’s data center business, and I will go over a live demo and explain the solution in more detail.