When releasing a new major software upgrade, publishers generally move up the major version number to the next available integer (eg Microsoft moving from Windows 10 to Windows 11). At Quividi, we must be loving the lucky figure 7, since we decided to name our most recent version of VidiReports: 7.7.
This new version is still operating indeed in the new architecture that we came up with in 2021 and called our 7th version, but it is worth stopping a moment to take a look at the various technical choices we made in anticipation of new market dynamics, and which are fully revealed in this version 7.7.
1/ The power of a modular, unified architecture
When we rewrote VidiReports from the ground up in 2018-2019, we got inspired by the automotive industry, which focuses a lot of energy on designing a modular and flexible chassis that can accommodate a whole range of car models addressing various market needs.
We built VidiReports to be our universal platform: it can run a wide range of computer vision algorithms in parallel, it shares audience data in real time via open APIs and it ensures that the data is securely and reliably uploaded to the cloud. VidiReports can run on an ever growing range of computing platforms and it is the only audience analytics solution today that can be installed on PCs powered by a $10 processor as well as on 64-core servers, at optimal cost.
The enduring success of Quividi’s approach is :
- our focus since 2006 to deliver a rock-solid analytics engine that consistently delivers audience data independently of computing platforms and venue type; audience data is produced within 5 minutes of initial set up, with no complex installation steps or need for calibration
- our ability to leverage today’s best-in-class computer vision algorithms, vetted and optimized for the available CPU; our open framework makes it possible to integrate new technologies to our released product in a matter of weeks ;
- our pledge to privacy-by-design principles, and our understanding of privacy protection legislations at the international level
- our native integration with over 40 digital signage systems (such as CMS and Adtech platforms) with which we automate data exchange – see the whole list here.
2/ The advent of neural networks
Neural networks are arguably today’s main driving force in computer vision but, in general, designing training a neural network from scratch is a resource-hungry and very specialized task that is best left to specialized teams. At Quividi, we decided to focus on what we do best, namely provide our customers with reliability and data-driven insights, while leveraging the excellence of choice partners for our deep learning modules:
- on Intel Architectures, our inference engine of choice is Intel’s own OpenVINO, fueled by both proprietary and open neural networks for the detection of people and vehicles, and for object classification. OpenVINO offers excellent performance at long detection distances and runs optimally on recent Intel processors.
- on ARM, where resources are severely limited, our newly announced partnership with Plumerai allowed us to introduce binarized CNNs that can count impressions in real time, at long distances, with precise dwell time, on mid-range, affordable players.
3/ Addressing Instore Communication at large
Quividi has built a solid expertise in retail media, working with leading convenience and grocery stores, retail kiosks operators, EV Charging companies and, of course, shopping malls, across five continents.
Media owners in retail use our high-fidelity audience impressions to monetize their screen networks (directly or, more recently, programmatically) while leveraging our real-time shopper engagement data to drive sales up.
VidiReports 7.7 further expands our in-store communication capabilities, as our platform can now be fully used in smart machines (vending machines, coolers…), digital merchandising units or retail shelf screens, bringing new opportunities to analyze, monetize and engage shoppers along the physical consumer journey.
4/ The digitization of downtown
In urban centers, we’re witnessing more and more 75’’ screens offering community services supported by ads: bus shelters, EV charging stations, citizen information panels, etc.
For those systems to thrive, in a world of programmatic and cookie-free targeting, solutions like VidiReports’ car and footfall real-time counting are ones that help those operators demonstrate solid and proven audiences, and offer value-added scenarios for advertisers.
In its latest 7.7 release, VidiReports can now detect people and vehicles up to 30% further.
5/ VidiCenter, in the shadow of VidiReports
VidiReports would not have prospered without its companion back-office platform. VidiCenter not only securely collects all audience and monitoring anonymous data from the tens of thousands of instances of VidiReports worldwide, but also offers a variety of services for our users, such as:
- multiple ways to expose the data: in one-click dashboards, in CSV exports, in open APIs, in automated workflows that push it to programmatic platforms;
- insightful technical info, to gauge whether a location is performing well, and benchmark it with others;
maintaining the instances up-to-date with fine upgrade mechanisms and reallocating license credits across new instances.
We see VidiReports 7.7 as the culmination of over a decade of smart technical decisions that are in tune with the evolution of the digital signage market. We can’t wait to get you to give it a try and see for yourself!
For more technical details on VidiReports 7.7, please go to this article in our Knowledge Base.