We created Neurolabs to solve the problem of high costs and manual labour involved in annotating everyday objects for image recognition (IR) in machine learning recognisers. Our breakthrough came from recognising real-world objects with high accuracy by training computer models with millions of synthetic scene versions and simulating real-world conditions.
We swiftly recognised that applying digital twins enabled by Synthetic Image Recognition (SIR) could be used beyond object recognition. These digital replicas proved versatile, extending their utility from integration into virtual environments to creating photorealistic 3D images for e-commerce platforms.
The Grocery and Consumer Packaged Goods (CPG) sectors were among the early adopters, captivated by the transformative potential of our technology. Working with leading Global brands, we developed the idea of creating digital twins that mirrored a product's lifecycle. By having access to digital twins across the product's lifecycle, clients could see how they could materially improve their operations.
In this blog post, we will outline the lifecycle of a typical product and explore how a CPG can benefit and improve its operations across that lifecycle, from manufacturing to recycling.
Digital Twins
Digital Twins were first voiced in 1991 in a publication by Gelernter called Mirror Worlds, but their first known application in manufacturing was by Dr. Grieves in 2002. The first applications were creating simulations to model an object's behaviour before manufacturing a prototype. Still, with the rise of the Internet of Things (IoT), applications rapidly expanded to measuring the output of sensors on real-world objects and managing the real world.
The traditional view of a "Digital Twin" refers to a digital representation or virtual counterpart of a physical object, system, or process. A “twin” is a digital model replicating the real-world entity in a digital environment, allowing for real-time monitoring, analysis, and simulation.
The concept involves creating a dynamic and interconnected digital replica that mirrors the physical entity's characteristics, behaviour, and changes over time. By collecting and integrating data from sensors, IoT devices, and other sources, Digital Twins enable organisations to gain insights into the performance, condition, and status of the corresponding physical object or system.
At Neurolabs, we believe that, like many advanced users of Digital Twins such as Rolls Royce, Digital Twins are multi-layered and are of most value when context is included in the twin.
What do we mean by context? A context could be a process that is applied to an object. Commonly, one would think of manufacturing, but at Neurolabs, we would, for example, add the context of Marketing. Imagine a SIR-enabled sensor identifying a promotional version of product packaging that is now being manufactured and alerting the Marketing team to initiate the planned marketing activities. This version of the digital twin exists in parallel with the manufacturing “twin” that doesn’t care about the packaging version. The manufacturing twin's purpose is to monitor the weight, shape, and volume to ensure the manufacturing line is running smoothly. In both cases, there is a “model” of the packaging, the latter recognising both special “promotional” and “regular” as the same item. Similarly, at “end of life”, another version of the twin would be the deformed packaging that allows SIR-sensors to differentiate between recyclable products and general waste.
Historically, it was expensive to create digital twins, so the "twinning" was initially limited to industries with high-value products. However, more and more companies are using the technique to improve their businesses. At Neurolabs, we believe the time is already upon us to apply the concept to everyday products on grocery store shelves everywhere. Our digital twin approach has multiple attributes, such as:
- Fully rendered packaging including deformed packaging
- Unique ID of each product version
- Nutritional information
- Weight, dimensions, volume, location
- Inclusion of EAN/UPC barcodes
This wealth of data provides comprehensive insights across all aspects of CPG’s operations. By embracing the multi-layered nature of Digital Twins and integrating context, we aim to redefine their application, offering a transformative impact on the retail industry.
What are CPG's priorities?
Many consulting organisations have surveyed CPGs and synthesised what they consider essential to the success of their business. Those essentials have stayed the same over time. They are to manufacture, distribute, sell, and increasingly sustainably recycle their products with minimal waste of raw materials at a healthy profit.
However, in an increasingly competitive market, the difference between those that execute well and those that do not is stark. McKinsey reports that while every company "leaks" value at various stages in their process, those who follow through and sustain their initiatives retain 2.0x the value compared to those whose "peter out."
In recent studies, both McKinsey and Salesforce highlight that achieving step changes in:
- Functional Excellence
- Adapting to the Modern Consumer
- Increasing Enterprise Resilience
are the keys to staying ahead. In short, it is all about persistently transforming business processes to meet market challenges.
Achieving this transformation is extremely difficult without data about the product at every process step. At present, gaps in data across the life cycle mean that collaboration and process improvements are hindered or even unimplementable.
To illustrate how access to that missing information can enable change, improve collaboration and increase value, we looked at the life cycle of a product (SKU) and where our capability could unlock the bottlenecks.
Manufacturing
Whilst we have some experience with products before packaging, see our work with Sagra Technologies, SIR shines post-packaging. The technology is straightforwardly used to:
- Verify labels to ensure a product does not leave manufacturing with incorrect labelling, such as wrong batch and expiry dates, avoiding expensive recalls in the supply chain. In the US, the Food Marketing Institute found that recalls cost companies an average of $10 million in direct costs alone.
- Identifying specific promotional packaging in the manufacturing process, enabling collaboration with marketing to trigger campaign execution precisely. According to Salesforce research, 87% of CPGs spend between 20% and 50% of gross revenue on promotion, so getting the timing right is revolutionary.
Logistics
As we now have fully rendered models of our product, we can start to do clever scheduling of promotions from within automated warehouses. In an automated warehouse, a picking robot such as those used by Ocado could be used to choose a specific product version to send to a particular post-code or for timed delivery to coincide with external trade promotions or digital marketing.
Similarly, if a specific product is in a local quick commerce centre, SIR-enabled picking devices can pick a precise product with the same benefits.
SIR-enabled logistics makes it straightforward to track exact products arriving at goods inwards and equally at goods out, something a barcode cannot replicate.
Further, as mentioned in the manufacturing section, if a product recall is initiated, those robots or SIR-enabled picking devices could automatically remove the product from the warehouse stock.
Retail
Once a specific product arrives at a retail location, SIR begins to shine. To set some context, Trade promotion, according to McKinsey, is up to 20% of a CPG’s revenue, and by extension, poor delivery of the process from budgeting through execution to evaluation has a material impact on the bottom line.
As the marketing team has been notified that a particular variant is now in the logistics chain, one can more precisely coordinate digital marketing campaigns to drive footfall, making trade promotions more effective, a capability unheard of previously.
With this knowledge, field teams can be timetabled to encourage compliance using SIR-enabled devices. Since compliance has been estimated to be 55% by the Promotion Optimisation Institute, gains in compliance will revolutionise the industry.
For all the focus of a CPG to improve promotion effectiveness, a customer can only buy what is on the shelf! Given that it’s known that 65% of shoppers don't buy the product when presented with an out-of-stock (OOS), it is imperative that shelves for those items, which, as a CPG, you have paid to be promoted, are replenished promptly. With cameras (fixed or mobile) equipped with SIR, OOS can be recognised in real-time, and rapid action can be taken.
With the stress driven by the cost of living crisis, Grocers are experiencing far higher shrinkage levels. With the rise of self-scan, a typical behaviour is to scan the barcode of a less expensive item and place a more expensive item in the cart. “Swapping” has now extended to produce. With SIR recognition cameras placed above weight scales, shoppers can be nudged to identify the item correctly. The same capability for packaged goods can be enabled at self-checkout, reducing shrinkage.
Continuing on the theme of loss, another increasing cause of revenue loss is the rise of counterfeiting. Initially, counterfeits were confined to high-value items such as designer clothing, but counterfeits are increasingly popping up at lower price points. Counterfeit premium alcohol is sufficiently widespread to account for losses totalling $500-$700m/pa. Often, the counterfeits have subtlety different packaging, with SIR-detection embedded watermarks, fakes can be detected quickly and cheaply.
It is an urban legend that Cashiers at certain supermarkets are very quick at processing a shop. The cashier's skill is to orient a product so the till immediately scans a barcode. With SIR-enabled tills, orientation isn’t a problem, enabling more rapid checkouts with less staff per open lane.
In the Home
According to Wikipedia, LG released the first Internet-connected fridge in 2000, which used cameras to “know” what was inside. However, since then, with the rise of the smartphone, people now routinely scan a product to capture its nutritional value in a calorie-counting app or to add it to an online shopping order.
With SIR on those different apps, CPGs could pay to receive the information and build up a “profile” of a household or pay to or offer a promotion to incent the shopper to add a substitute to the order.
Recycling
At the end of a product's lifecycle, there is almost always waste. CPGs, to keep perishable items fresh, have used packaging that continues to contribute to the vast mountains of waste plastic dumped into landfills. A crucial part of recycling is sorting the waste at the first disposal point. Not too long ago, humans were the only sorting devices. However, with the increasing rise in smart bins, beginning in public spaces, SIR-enabled bins will soon efficiently sort waste with models supplied by CPGs as a public service. The data collected enables the tracking of where and what can be recycled.
Sources:
Secrets to implementation success, MkKinsey 2016Operations as a competitive advantage in a disruptive environment, McKinsey 2016Consumer Goods Industry Insights Report, Salesforce 2023
At Neurolabs, we are revolutionising in-store retail performance with our advanced image recognition technology, ZIA. Our cutting-edge technology enables retailers, field marketing agencies and CPG brands to optimise store execution, enhance the customer experience, and boost revenue as we build the most comprehensive 3D asset library for product recognition in the CPG industry.
