Last month, we attended the O’Reilly AI Conference in London, UK. This conference series brings together cutting-edge technology companies and a diverse audience of technical and business leaders to share knowledge and ideas on Artificial Intelligence (AI).
The event featured high-level keynotes, technical presentations, training sessions and demonstrations of AI solutions. Some example applications included:
- AWS Deep Racer – a re-enforcement learning tool that simulates car races encouraging players to optimise their learning model to win the race, powered by an Intel® Atom™ processor and optimized by the Intel® Distribution of OpenVINO™ toolkit
- Gramener – Predicting quality of life using satellite data and deep learning
- Bluewhale – Measuring quality of large datasets using anomaly detection
We were excited to join the event and showcase our own computer vision demonstrator developed in partnership with Intel.
Our Visual Object Identifier Demonstrator – or VOID – is a real-time person identifier that locates people and detects some of their physical features. It calculates real-time locations and provides estimates of age, gender, emotion, head orientation and clothes. It works with two or more cameras and is smart enough to identify the same person on different images. It was hugely enabled by the Intel® Distribution of OpenVINO™ toolkit that provides pretrained computer vision models that can be quickly assembled into image processing pipelines for Intel® processors. In addition, the toolkit optimizes deep learning deployment across multiple Intel® architecture, including CPU, integrated graphics, VPU and FPGA.
Applications of the solution include improving a retail store’s awareness of its customers. VOID generates real-time data that, tied together with point-of-sale or digital signage data, provides behavioural insights that can help retailers better engage with their customers and improve their business outcomes.
During the conference, we ran the demo using 4 cameras overlooking the exhibition space. Attendees coming to our stand could see their location and face features plotted in real-time on a dashboard screen.
As members of the Intel® AI Builders program, we were able to showcase our solution in a high-profile AI context and to share it with other inspiring partners showing a range of how Intel® AI technology can be applied. Solutions like our real-time tracker, voice assistant and visual product recognition shown together in one space gave a tantalising glimpse of AI connected experiences that will be part of the future of retail.
Check our other retail & data science stories on our website: www.valtech.com