AI, AR, IR – terms that seem both very impressive and very complicated. However, while a lot of science and calculations goes into making them work, the whole point of using this type of technology is to make your work/life a million times easier.
So let’s talk image recognition: a technology that’s making it possible for CPG field sales teams to complete their store audits in less than half the time of manual methods, while still maintaining accuracy and even boosting efficiency.
But how?
Whenever something seems too good to be true, it’s always good to take a peak behind to curtain to see what’s happening behind the scenes.
In this blog, you’re going to do exactly that and see how BeMyEye’s image recognition tech allows your retail audits to happen in the blink of an eye.
1. Snap, snap, snap again
Shelves come in all different shapes and sizes, which means it’s highly unlikely that your reps can capture the full scope of a shelf with just one picture.
In order for IR to work as intended, it requires a complete visual. To get all the data necessary, reps will need to follow an on-screen guide and take several pictures of the shelf that will then feed into the magic algorithm known as the neural network.
2. Let's get stitchy
Once all the pictures have been taken by your rep, they’re shipped over to the server to be stitched together into one complete visual.
The AI does this by recognising similar features between the photos and placing them together in a mosaic. This is where the image recognition kicks in, as it makes sure that there’s no overlap from the pictures and that items aren’t repeated.
For example, if a canned drink is to the far right of one picture and the same drink to the far left of another, these can be lined up and stitched together so that the drink only shows up once in your results.
The final stitched together picture is classed as linear panorama.
3. Hey, I recognise you!
To make sure the accuracy of the image detection is as good as it can get, the stitched shelf picture is then segmented up into a grid.
The AI scans each quadrant in two stages: at a product level and at a SKU level.
Unlike your rep taking the picture, the AI has no prior knowledge of the category that it’s about to be presented with. This means that at first, it needs to scan the picture at a very basic level so that the correct algorithms line up for the next step.
This could be as simple as identifying all the body care products in a picture of a supermarket shelf. Anything featured in the photo that isn’t part of the correct category will be ignored.
After this, the recognition will narrow its analysis down to specific products and SKUs – the exact intelligence that you’re looking for.
4. Look at all those insights
Although they seem complicated to us humans, the AI that performs these tasks makes it look easy, completing the three steps in less than a second.
Once that’s done, the results are accuracy checked by either the algorithm itself (this is how real-time image recognition is achieved) or a specialist on our team. In the former, you get your KPIs in just a few seconds, and with human help, it can take anywhere from a few minutes to a few hours.
At the end of the process, you’ll receive all your store KPIs and your reps can quickly move onto what they do best – selling! It’s a win-win for everyone.
If you’d like to learn more about how real-time IR is achieved and how your team can use it to complete more efficient stores audits, read the full outlook in our white paper “SCANNING THE SHELVES – How Image Recognition Helps Consumer Brands To Boost Their Retail Execution” here.
