You’re heading home after a busy day at the office, and realize that there’s nothing in your kitchen for dinner. You enter the store, and don’t have a clue what to buy. But what if someone could tell what you really wanted just by looking at your face? Intel and Kraft are saying that’s exactly what their new store kiosk, the “Meal Planning Solution,” will do.
They claim that the computer will be able to scan your face and determine what kind of dishes you might like. It will also keep a record of your purchase history in the store and any history you might have on Kraft’s internet site or their iFood Assistant mobile app. We dug a little bit into the science behind this technology – here’s what we found. At first listen, the thought of having a computer look at your face and then spit out a shopping list for you seems bizarre. But it turns out that Intel and Kraft have been working together since 2010 to come up with something that actually creates solutions to four different groups:
Shoppers
People like the idea of making a grocery list, but tend not to follow through. Nearly three-quarters of the time, they go shopping without a clear idea of what they plan to cook, and so don’t know what to buy. To top it off, most folks only have ten or so meals that they know how to cook well and that the whole family likes. So how do we combine all this information to allow us to help people create decent meal plans and shopping lists?
Stores
For retailers, it’s all about the bottom line – and having a service that no other store provides helps increase both first time and repeat traffic in the store. Additionally, making recommendations increases purchases.
Kraft
Obviously, their goal is to sell more products. They have an excellent presence online and have a department developing recipes that use their products. With the in store Meal Planner, Kraft is able to get their recipes to people who need them.
Intel
They’ve been working on video analytics for years, and it’s important for them to be able to find partners to work with who can put their technology to work in a way that helps the end user. What makes the Meal Planning Solution such an innovative solution is its use of software that uses anonymous video analytics to determine age and gender, and with that information it’s able to make reasonable suggestions for recipes. The software has been shown to be from 70-80% accurate in its determinations. The sensors work by scanning the face and sending the information to the software. The first thing the software does is confirm that the image is actually the pattern of a face, by looking for certain tell-tale signs: horizontally matched shadows in the eye region, a brightness in the nose position and so on. Once it determines that it is actually a human face, it moves on to the rest of the algorithms.
The face detector algorithms are trained to recognize faces by scanning thousands of pictures that are maintained on a database. It compares similar features to determine if it has scanned a face that belongs to a man or a woman, and then separates the pictures into age brackets – child, youth, adult, senior. The scan data is processed for the necessary information, and then destroyed; none of the images are stored. The computer also tracks the number of people who use the Solution Center at a given site, and keeps track of how much time each person spends at the kiosk. So next time you're stuck for something to cook, let a computer see if it can give you some hints!
Jess has always been fascinated by technology. Combine this with her love of food, and she has the perfect job at RJ Herbert as an online researcher and blogger for the vegetable and food processing company.
They claim that the computer will be able to scan your face and determine what kind of dishes you might like. It will also keep a record of your purchase history in the store and any history you might have on Kraft’s internet site or their iFood Assistant mobile app. We dug a little bit into the science behind this technology – here’s what we found. At first listen, the thought of having a computer look at your face and then spit out a shopping list for you seems bizarre. But it turns out that Intel and Kraft have been working together since 2010 to come up with something that actually creates solutions to four different groups:
Shoppers
People like the idea of making a grocery list, but tend not to follow through. Nearly three-quarters of the time, they go shopping without a clear idea of what they plan to cook, and so don’t know what to buy. To top it off, most folks only have ten or so meals that they know how to cook well and that the whole family likes. So how do we combine all this information to allow us to help people create decent meal plans and shopping lists?
Stores
For retailers, it’s all about the bottom line – and having a service that no other store provides helps increase both first time and repeat traffic in the store. Additionally, making recommendations increases purchases.
Kraft
Obviously, their goal is to sell more products. They have an excellent presence online and have a department developing recipes that use their products. With the in store Meal Planner, Kraft is able to get their recipes to people who need them.
Intel
They’ve been working on video analytics for years, and it’s important for them to be able to find partners to work with who can put their technology to work in a way that helps the end user. What makes the Meal Planning Solution such an innovative solution is its use of software that uses anonymous video analytics to determine age and gender, and with that information it’s able to make reasonable suggestions for recipes. The software has been shown to be from 70-80% accurate in its determinations. The sensors work by scanning the face and sending the information to the software. The first thing the software does is confirm that the image is actually the pattern of a face, by looking for certain tell-tale signs: horizontally matched shadows in the eye region, a brightness in the nose position and so on. Once it determines that it is actually a human face, it moves on to the rest of the algorithms.
The face detector algorithms are trained to recognize faces by scanning thousands of pictures that are maintained on a database. It compares similar features to determine if it has scanned a face that belongs to a man or a woman, and then separates the pictures into age brackets – child, youth, adult, senior. The scan data is processed for the necessary information, and then destroyed; none of the images are stored. The computer also tracks the number of people who use the Solution Center at a given site, and keeps track of how much time each person spends at the kiosk. So next time you're stuck for something to cook, let a computer see if it can give you some hints!
Jess has always been fascinated by technology. Combine this with her love of food, and she has the perfect job at RJ Herbert as an online researcher and blogger for the vegetable and food processing company.