How AI Will Shape the Gym of the Future
Amazon Go is one of the most notable examples of how computer visioning and sensor data is transforming our world. By offering an in-store experience without checkout lines, Amazon claims that it has created "the world's most advanced shopping technology." The underlying concept can be applied well beyond food retail and has the potential to revolutionize many industries.
When applied to the fitness and health industry, advanced IoT and computer vision technology can help accelerate the path toward a truly smart gym. These solutions can track an individual’s in-gym workout to provide a complete representation of the activity data required for AI-enabled digital fitness and health coaches. Similar to how Amazon Go automatically tracks the content of the customer's grocery shopping, a computer vision-supported workout tracking system allows users to determine the exact exercises and repetitions, as well as the intensity and quality of exercise execution of any given workout.
The Gym of the Future is an ecosystem that provides a differentiated opportunity for health club chains by allowing health club members to automatically track their workout progress with a full range of exercises.
Giving health club members the option to access a solution that automatically tracks their workout routines across the full spectrum of gym activities gives them unprecedented transparency on progress and provides tailored recommendations. On top of that, health club chains gain unique data insights. Used in the right form, this data will generate added value for both customers and the companies in areas ranging from customer engagement to internal operations.
AI-enabled Applications Drive the Need for Better Activity Tracking
Many startups and established companies in the fitness and health industry are investing in AI-enabled personalized coaching, health and nutrition apps to help people stay active, healthy and advance their workout routines. For the most part, those applications use AI-driven recommendation engines that use collected data to improve suggestions for food, workouts, or other feedback and coaching.
Getting reliable activity data is one of the critical ingredients for such applications. Many apps rely on user input to capture this essential information, leading to data quality that is highly dependent on the individual, making it difficult for even the smartest AI coach to come up with meaningful recommendations. Activity trackers, like Fitbit or the Apple Watch, have improved the reliability and amount of tracked activity data, and fitness equipment providers are contributing to the data stream by offering increasingly more sophisticated connected devices to health clubs. However, activity trackers by design face limitations in what they are able to track and IoT fitness equipment is, for the most part, constrained to measuring only the single activity they are designed for. This is why we still see people using the old-school, pen-and-paper method for tracking their workouts.
Adding Computer Visioning to the Mix
Computer vision technology as an additional data stream that recognizes specific exercises, exercise durations, and number of repetitions or weights used will capture data that is otherwise hard to track, elevating exercise tracking to a new level. A combined system of computer vision and IoT sensors enables a full log of the workout routine as well as detailed statistics that involves no human intervention.
The Gym of the Future at EPAM
Engineering such a system is certainly not trivial. Machine learning pipelines for human activity recognition are typically rather complex and involve multiple processing steps. Furthermore, the computer vision system needs to be trained to detect a wide range of special objects to recognize various equipment.
EPAM Garage’s Gym of the Future proof of concept demonstrated that a small team of experienced engineers can achieve remarkable results in just a few weeks and shows the feasibility of one of these systems. The below video shows the results of the first iteration of the concept:
The technologies required for a smart gym are already available today, and multiple startups are launching computer vision-based applications in the sports and fitness industries. It’s only a matter of time until a truly holistic solution will exist. Health clubs are uniquely positioned to drive detailed activity tracking for their customers for two main reasons. First, for many people, gym activities are an essential and sometimes large part of their workout routines and consequently, tracking this data is extremely valuable for gym customers. Secondly, health clubs are in the prime position to track this data, as they have a fully controllable environment. Evolving a Gym of the Future system thus offers tremendous value for all parties and sets us on the path for an entirely virtual coach.