How Artificial Intelligence Is Changing The Fitness Industry During COVID-19

The fitness industry is undergoing a major transformation through a massive deployment of IoT applications and innovative artificial intelligence (AI) product offerings. The research firm Reports and Data predicts that the annual revenues for the fitness app market will reach $14.64 billion by 2027 with around 100.2 mln of fitness app users by 2024.

AI-powered applications in the health, nutritional, and fitness sectors are finding exceptional consumer demand. Hiring an AI engineer, startup owners are creating smarter products, leveraging the latest innovations in machine learning, deep learning, and computer vision.

AI-Based Personal Trainer

Visiting a gym becomes risky or even impossible with local lockdowns in the time of COVID-19, so that AI-based personal trainers are seeing a noticeable surge in popularity.
Physical trainers know that maintaining good posture is a critical element in any fitness routine. Doing some exercises with bad posture can cause harm or injury to the person executing the move incorrectly. AI-driven human pose estimation technology is one innovation that makes AI personal trainer apps exciting.

Human Pose Estimation Technology

Human pose estimation is an advanced computer vision-based technology. It is similar to facial recognition but for the entire body. Human pose estimation systems detect and analyze the human body’s position by three analytical methods, which are:

• Modeling the Skeleton: This uses key points to represent the skeletal structure of the human body.
• Modeling Contours: This uses the rough width of the body’s torso and limbs to show rectangular boundary boxes of a person’s silhouette.
• Modeling Volume: This analytical method uses the capture of 3D body scans and then represents the body using geometric meshes and shapes.

Zenia and Freeletics are examples of fitness applications that make use of AI-driven innovation.

Zenia is an AI-powered digital assistant for those practicing yoga. To do yoga correctly, it is critical to know you are practicing the positions with the right posture. In-person yoga instructors are constantly helping students to correct their posture and to hold each yoga position properly.

Zenia works by having a new user download and install the app on a smartphone with a camera, set up an account, and place the smartphone camera with a full-front view of the area where the yoga practices will be done.

The app recognizes 16 joints of the body and uses voice assistance to guide the user. The AI component also adapts to each practitioner’s pace and monitors progress while giving feedback about improvements made.

Zenia used a training set of 200,000 yoga poses, some done correctly and others incorrectly, to teach the software through machine learning. The software is able to recognize when a student is doing something wrong. Professional yoga instructors were also consulted during the software algorithms development.

To protect users’ privacy, all the AI calculations using the camera images happen on the user’s smartphone with edge computing software. None of the user’s session data is stored in a centralized location. The private components stay on the smartphone.

Freeletics is an AI-driven fitness app that Forbes says is used by over 47 million people in over 160 countries. The secret to Freeletics massive popularity is that the program creates a unique workout for each user that is presented from over 3.5 million options. Each customized program is designed to promote an individual’s health.

The app uses casual inference modeling combined with data mining to predict what a new user will be able to do by analyzing patterns from over 50 million other users. It modifies workouts by learning how a user reacts. This helps to address deficiencies in an exercise program and correct imbalances in the body’s development. The software adjusts workouts based on the user’s available equipment.

AI-driven fitness training may produce better results than having a human trainer because the software knows more exercises, has access to more data, and tracks progress more precisely.

Smart Clothes and Wearables

Another sector where AI-driven solutions are making rapid advancements is in smart clothes and wearables. Clothes and other wearable items contain sensors that connect to systems to help correct biomechanics (such as improving a golfer’s swing), improve performance metrics for athletes, and increase fitness. Asensei, Sensoria, and Wearable X (Nadi X brand) are three excellent examples in this category.

Asensei created a line of smart apparel, which includes black compression shirts and pants with five inertial sensors in each one. These sensors collect data that is useful for improving posture, technique, form, and timing when doing exercises and playing sports. The apparel and wearable bands containing sensors help improve strength training workouts and yoga. They have many uses in sports programs, especially those sports conducted in large spaces where using detailed camera coverage is challenging such as on golf courses and in baseball outfields.

Sensoria makes IoT-connected smart garments and socks. The upper worn garments measure heart rate. The socks measure cadence, foot landing patterns, and impact forces. The Sensoria Run app tracks the heartbeat of a person while jogging or running and can tell how far they run and how fast. The company also offers a software developers’ kit that encourages other developers to expand the uses of their products with AI-driven apps that interface with the data collected from the smart apparel.

Wearable X makes the line of Nadi X yoga pants that contain two types of sensors. The sensors are integrated accelerometers and haptic monitors. Haptic monitors create vibrations by spinning a mass that reacts to movements like a small gyroscope. The sensors detect positioning on the X-, Y-, and Z-axis in relation to the earth’s gravity field.

There is a battery-operated microcontroller that sends data via a Bluetooth wireless signal to an app on a nearby smartphone or Bluetooth-enabled device. The batteries are lithium polymer and rechargeable by using a micro-USB connection.

AI-Driven Diet Planning

AI-driven apps provide motivation and support with nutritional meal-planning and dieting efforts. Two popular apps in this category are FitGenie and HealthifyMe.

FitGenie describes itself as a smart calorie-counter. It works by using a machine-learning algorithm to make suggestions for nutritional planning. This is motivational support for those who want to reach a certain weight while dieting or to achieve a specific fitness goal.

The algorithm maps and forecasts an individual’s progress and then makes weekly adjustments based on the information collected. The data modeling works with scientific data, experience-based learning from human life-coaches, body composition data, diet plan adherence, rate of weight change, hunger, fatigue, and other customized metrics. The algorithm constantly recalculates based on all of the parameters to make informed adjustments and suggestions.

HealthifyMe was developed in India. It cleverly uses an AI-powered virtual nutritionist character called Ria. Ria helps users with questions they have about fitness and nutrition. It works with audio and text inquiries in more than 10 languages.
The data used for machine learning includes over 250 million tracked foods, workouts, and more than 10 million messages exchanged between users and human nutritional coaches. The system works with wearable devices that are popular in India.

Conclusion

Being an emerging trend, AI-powered applications in the fitness industry still face some technological limitations. Data science and machine learning engineers are experimenting with new approaches and leveraging the latest innovations.