RESEARCH LAB
Fitbyte
Changes for the better
Most of us must have faced the insoluble dilemma of whether or not having a donut while being on a diet. ‘A bite won’t hurt’ and ‘My dietician would not even know!’ are some of the common reassurances we give ourselves. Regardless, would being oblivious to the food intake really assist the diet? How would you assess the impact of a questionable convenience food item on your health? Wouldn’t it be well favored if your mobile phones would keep you posted about your food habits?
The prodigious invention of a ‘wearable’ device capable of tracking the eating habits with high precision, has proven to be noteworthy for its affirmative results. Such innovations would never fail to amuse people because us conventional humans have always been intrigued and inspired by the idea of a digital medium being made to learn the ropes of a human! “Whether it is Google, or Apple or free software, we have got some fantastic competitors and it keeps us on our toes.” is rightly emphasized by Bill Gates, inclining people more towards contributing and ensuring the exponential growth of technologies.
The Institute for Software Research of Carnegie Mellon University, a private research university based in Pittsburgh, Pennsylvania, USA stands among the world’s most renowned educational institutions with its cutting-edge brain science, path breaking performances, and innovative start-ups. Institute of Software Research being one of the seven academic departments of the Carnegie Mellon school of Computer Science, focuses on research pertaining to software engineering, privacy and security engineering, and societal computing.
‘Fitbyte’ as named by the researchers is a wearable sensing system that can be attached to any pair of consumer eyeglasses. Combining the detection of sound, vibration and movement, the system assures increased accuracy and decreases false positives. The multimodal disposition of the system helps it detect chewing, swallowing, hand to mouth gestures, and also, visuals of intake. The gadget has multiple integrated IMU (Inertial Measurement Unit) sensors with a gyroscope and an accelerometer. A side of the apparatus has an equipped camera, while the other has a proximity sensor. “The primary sensors on the device are accelerometers and gyroscopes, which are in almost every device at this point, like your phones and watches.” says Mayank Goel an assistant professor in the Institute of Software Research, thereby stating the regularity of devices with fitted sensors.
The infrared proximity sensor detects hand to mouth gestures. To spot any kind of chewing action, the system monitors jaw motion using four gyroscopes around the wearer’s ears. By viewing the area behind the ear, the system succeeds to trace the flexing of the temporal muscles as the user continues to move their jaw. High speed accelerometers placed near the glasses’ earpiece perceive throat vibrations during swallowing. The technology addresses the longstanding challenge of inch-perfectly detecting drinking and the intake of soft food or drinks like yogurt. With the vivid graphical outputs from the accelerometer, proximity sensors and gyroscope, the system analyses the user’s activity by contemplating any ebbs and flows in the produced results. The small camera fitted on the side, superintended for gathering the visuals, points downwards to capture just the area around the mouth and only turns on when the model detects the user eating or drinking. As in regard to addressing issues of privacy, the captured images are not shared anywhere except the user’s registered phone.
As the system is dependent on the photos to recognize the food and drinks consumed by the user, the research team has plans for incorporating datasets with Deep Learning modules which will assist to determine the food type. The photos could be then uploaded on the user’s mobile application providing the details about the time and type of intake. The team affirms to install additional sensors to allow the model to detect blood glucose levels and other physiological measures. Fitbyte has been currently tested in five unconstrained situations including a lunch meeting, watching TV, having a quick snack, exercising in a gym and hiking outdoors. Given the proposed future scope inclusions, the gadget is predicted to have the ability of whole handedly managing to captivate maximum audience.
The customary diet monitoring method being journaling, relies heavily on self-reports that could suffer from low adherence and recall errors. Though not being wholly deployed currently, Fitbyte poses to be a potential alternative to any other diet monitoring methods sealing errors caused by human analysis. Seems like our dietician can now incontestably ‘see’ what we eat, and having a donut can no longer be a stealth mission!
References: https://www.cmu.edu/news/stories/archives/2020/april/fitbyte-monitors-diet.html
