Qopius: Delivering innovative digital inventory insights to brick-and-mortar retailers

About Qopius

Qopius helps retailers to digitize brick-and-mortar retail by offering reliable, real-time shelf analysis using hardware-agnostic computer vision.

Industries: Technology
Location: France

Qopius implements Google Cloud Platform to improve GPU stability by deploying on Compute Engine, improving system reliability, and using BigQuery to deliver customer insights in seconds.

Google Cloud Results

  • Improves system reliability by deploying GPUs on Google Compute Engine for a more stable infrastructure
  • Delivers customer insights in seconds by processing and visualizing data using Google BigQuery and Google Data Studio
  • Enables agile development with a more efficient workflow thanks to a unified work environment

100K products detected and analyzed so far

High street retailers may feel margins being squeezed, but innovative technologies can help to reduce operational costs. By providing brick-and-mortar retailers with a way to analyze their shelves in real time, French company Qopius offers them the digital tools they need to improve efficiency and optimize margins. "Our solution can detect stock that is out-of-date, or misplaced," explains Bedeir Rizk, CMO at Qopius. "It empowers the store managers and staff with the knowledge they need to improve their operations, and offers suggestions on what needs to be done." Qopius's customers include leading retailers such as MediaMarktSaturn, Albert Heijn, and Metro, as well as industrial leaders.

"One of our differentiators is that we see things from the retailers' perspective: our product has to answer their needs and be easy to use. Real-time analysis is key, so we need a cloud provider that's reliable and available."

Bedeir Rizk, CMO, Qopius

"We worked together with Qopius to digitize our stores. Using Qopius, we learned the layout of our store and the corresponding product location that helps us build a better customer experience in the future. The collaboration with Qopius was great. The team is very ambitious and we got excellent results for recognizing the products in our stores," says Martin Wild, Group CIO of MediaMarktSaturn.

Unlike leading competitors, Qopius' solution is hardware-agnostic and can be integrated in any store. "Since starting the business four years ago, we've detected and analyzed over 100,000 products," says Bedeir. "We collect shelf data via numerous sources: the solution can be integrated with existing hardware such as a security camera, and staff can also take photos via a mobile app."

Qopius uses GPUs to process large volumes of image data so it can detect, identify, and analyze products on the shelves. To handle demand from its growing customer base, it wanted to achieve greater stability, so it decided to switch from its previous cloud provider to Google Cloud Platform (GCP).

"One of our differentiators is that we see things from the retailers' perspective: our product has to answer their needs and be easy to use," says Bedeir. "Real-time analysis is key, so we need a cloud provider that's reliable and available."

Building a stable platform for GPU usage

Providing customers with real-time analysis requires a stable platform, especially if a business offers high availability as part of its customer guarantees. "With our previous provider, we were not able to achieve the levels of reliability we needed, due to downtimes and failed GPU tasks," says Bedeir.

To improve the stability of its infrastructure, Qopius migrated its processing to virtual machines on Compute Engine. "It took around six months to build the new architecture on GCP," explains Bedeir. "We wanted to migrate quickly, and add new clients and projects once we were up and running."

"We use BigQuery to generate the KPIs we offer to our clients, helping them to improve their margins. We can transform shelves into actionable insights in seconds. As it offers such fast processing, we can integrate and visualize more data for the client, ultimately reducing their end cost."

Bedeir Rizk, CMO, Qopius

For its Retail Shelf Tracker tool, Qopius uses virtual machines on Compute Engine, and App Engine and Cloud Storage for processing and storing data. It then generates dashboards with optimization tips for its clients using BigQuery and Google Data Studio. Stackdriver Monitoring provides system visibility and oversight, so it's easier to keep track of each resource within the infrastructure.

"We use BigQuery to generate the KPIs we offer to our clients, helping them to improve their margins," says Bedeir. "We can transform shelves into actionable insights in seconds. As it offers such fast processing, we can integrate and visualize more data for the client, ultimately reducing their end cost."

Working with BigQuery and Data Studio has also helped Qopius to be more flexible in its offering. "Thanks to Data Studio, we're able to segment the data in many ways, enabling us to offer a wider choice of KPIs within our dashboards," says Bedeir. "Our customers really appreciate our flexibility in tailoring the KPIs to their specific business needs, and Data Studio helps us to do that."

Store Dashboard

Accelerating developer workflow

One of the big advantages of GCP for Qopius is that it encourages an efficient workflow for its developers. "As all the Google Cloud Platform components are part of one environment, they communicate seamlessly," says Bedeir. "All the services are packaged together, without having to switch between interfaces, which speeds up the workflow for our developers. Managed services like Cloud Functions and Cloud Build help save time on the operations side, and all you need to connect to a virtual machine is your email address. A few seconds saved here and there all add up to a significant gain in overall efficiency."

Easy integration between GCP and other software also supports Qopius' machine learning development and software partnerships. "We use TensorFlow for some of our machine learning models, and we definitely benefit from the great support on GCP for that," says Bedeir. "We're also developing a partnership with SAP, and the existing partnership between SAP and Google Cloud will definitely facilitate integration between SAP software and our platform."

"In 2019, we'll be expanding operations with an additional seven retailers. That means many more images to process, so having a reliable infrastructure that's ready to scale is key to our strategic goals."

Bedeir Rizk, CMO, Qopius

Preparing for expansion

Since migrating to GCP, Qopius has gained the infrastructure stability it was looking for. "We've seen a real difference in the performance of our GPUs on Google Compute Engine," says Bedeir. "As processing large volumes of images is central to our business, that's a big relief."

Now, Qopius plans to extend its use of GCP tools, as well as expanding its client base. "The next step for us is to explore Google Kubernetes Engine, in order to automate further and improve our image processing time," says Bedeir.

"In 2019, we'll be expanding operations with an additional seven retailers," he explains. "That means many more images to process, so having a reliable infrastructure that's ready to scale is key to our strategic goals. Our aim is to be present in up to 100 stores in 2019, and with GCP, we have the support we need to achieve this."

About Qopius

Qopius helps retailers to digitize brick-and-mortar retail by offering reliable, real-time shelf analysis using hardware-agnostic computer vision.

Industries: Technology
Location: France