Google AI could keep baby food safe

 

Google/Kewpie Corporation

Google's artificial intelligence technology can help the food industry beyond picking better cucumbers. In one company's case, it could prevent your child from getting sick. Japanese food producer Kewpie Corporation has revealed that it's using Google's TensorFlow to quickly inspect ingredients, including the diced potatoes it uses in baby food. The firm and its partner BrainPad trained the machine learning system to recognize good ingredients by feeding it 18,000 photos, and set it to work looking for visual 'anomalies' that hint at sub-par potatoes. The result was an inspection system with "near-perfect" accuracy, culling more defective ingredients than humans alone -- even with a conveyor belt shuttling potatoes along at high speed.

Kewpie goes out of its way to stress that it's supplementing human workers, not replacing them. The TensorFlow system handles the initial removal, but there are still people waiting to inspect the potatoes afterward. Think of this as a booster that helps Kewpie process food faster without compromising quality.

The tech is currently limited to Kewpie's baby food, but the company expects to spread it to other foodstuffs before long. Also, it's not hogging the technology to itself. It wants to offer the AI-assisted inspections to other companies, so the industry at large might see a benefit. There's no guarantee that other companies will keep all their human staff (we could see others using this as a chance to cut jobs), but it could set a higher quality baseline.

تابع القراءه...

Soylent recalls its food bars after making some customers sick

 

You probably shouldn't eat one of Soylent's recently-launched food bars. The company is halting sales, and recalling existing bars following reports of of customers getting sick after eating its latest future-food. People have reported feeling nauseous, vomiting and even diarrhea.

"After hearing from our customers, we immediately began investigating the cause of the issue and whether it was linked to a problem with the Bars," the company said. "So far we have not yet identified one and this issue does not appear to affect our other drinks and powder. Though our investigation into this matter continues, we have decided to err on the side of caution and take this preventative step."

As reported by Ars Technica earlier this week, while many reports were sent directly to the company and its own official forum (it's not cheery morning reading), more appeared on Reddit threads, with complaints doubling since Monday.

Launched in August this year, (the same month that the smoking, exploding Galaxy Note 7 was also introduced to the world), Soylent has already removed the food bars from its online store for the time being and has cancelled shipping orders.

Mileage varies, but many customers suffered assorted tummy troubles a few hours after eating the bars. One unfortunate customer relays a story about having a reaction so severe that he felt faint, and had to be taken to hospital. Soylent has already tested the bars of some affected users for contaminated materials and bugs: those tests came back negative.

Some customers had eaten other Soylent products without problems, including the meal-replacement drink that brought the company to the attention of many -- especially Silicon Valley-based workers looking for a time-saving meal hack. The company says that each food bar packs 12.5 percent of your daily nutritional requirements — priced at two dollars per bar.

Ars was told by a source that the company was also looking into whether the soy-based nature of the bar was triggering food intolerance reactions. The soy protein inside each bar apparently comes from three different sources and they may not be mixed evenly as the bars are made, according to the source. Two of these aren't found in other Soylent products.

تابع القراءه...

Smart fabric can recognize the food you put on the table

 

Tatjana WAGNER via Getty Images

Wouldn’t it be helpful if a food tracking or recipe app could determine what you’re eating just by detecting what’s on your table? You might get your wish. Researchers at Microsoft and multiple universities have developed Capacitivo, a smart fabric system that can detect food, drinks, and other objects based solely on touch. The cloth uses a combination of a capacitive electrode grid and machine learning to gauge both the material and shape of a given item.

This works with some containers, too, including glasses and bowls.

The researchers initially wove the technology into a tablecloth, and envisioned it as particularly helpful for cooking. An app could suggest a meal based on what’s sitting on the table, while a diet app would know what you’ve had to drink. Scientists also imagine Capacitivo serving as a memory tool. It could remind you to take your earbuds before you head out the door, or remind you to clean up if you leave an empty food bowl.

The initial project has limits. It doesn’t recognize metallic objects, and won’t work as well with books or other square-edged items. It likewise has problems with credit cards and other items that don’t have a clear capacitance footprint. Certain drinks don’t produce reliable results, so you shouldn’t expect your tablecloth to tell an IPA from a stout.

Don’t expect a smart tablecloth in the near future, then, although the technology should get better over time. The team hopes to detect metal, deal with less-than-ideal placement and enable touch input, among other upgrades. If all goes well, it could lead to a truly seamless smart home where you don’t have to rely on smart scales, scanning or other clunky methods to detect what’s in your kitchen.

تابع القراءه...

Google hopes to count the calories in your food photos

 

Be careful about snapping pictures of your obscenely tasty meals -- one day, your phone might judge you for them. Google recently took the wraps off Im2Calories, a research project that uses deep learning algorithms to count the calories in food photos. The software spots the individual items on your plate and creates a final tally based on the calorie info available for those dishes. If it doesn't properly guess what you're eating, you can correct it yourself and improve the system over time. Ideally, Google will also draw from the collective wisdom of foodies to create a truly smart dietary tool -- enough experience and it could give you a solid estimate of how much energy you'll have to burn off at the gym.

This isn't going to lead to a practical product in the short term. Google only just filed for a patent on Im2Calories' underlying technology and has no immediate release plans, so you can post dessert photos to Instagram with relatively little guilt. Eventually, though, it could be a staple feature of health apps that help you balance your food habits with your activity levels. And the potential doesn't stop there, either. While food is the "killer app," the image recognition code could also apply to traffic prediction and anything else where a series of photos can provide a wealth of data.

[Image credit: Ana Arevalo/AFP/Getty Images]

تابع القراءه...