Line - Stay Injury Free

Overview of Activities
Injury Risk Profile and Muscle Load
Preventive Plan
Squat
Squat Movement
Detection of Correct Squat Execution

Inspiration

All of our team members do lots of sports - and unfortunately we get injured at some point. This is not only painful and expensive but it also limits our daily routines and performances. As we are already tracking our activity and sports data via smart watches, we came up with the idea of using this data to generate individual risk profiles with personalized recommendations for preventive exercises.

What it does

Line helps you to prevent injuries and avoid expensive and painful therapies by providing you with individual preventive measures. Connected to a Garmin smart watch, line collects comprehensive data from your daily activities, analysizes them and provides you with an overview of your current fitness level. Furthermore, it calculates your own risk profile of getting injured and hightlights if the current training is to intense or the context of the training is just not right. Based on these insights Line provides you with personalized suggestions, e. g. preventive exercises to reduce the injury risk. However, it is essential that these exercises are performed correctly in order to not worsen your body's condition. Therefore, Line helps you to gain proficiency in new movements by visually analyzing your movements and providing feedback. To focus on the relevant aspects only, your body is cropped and skeletal keypoints are displayed. Line tracks your progress along with other data, e. g. your heart rate which is directly retrieved from the Garmin watch.

How we built it

In the backend we used pytorch: (a) A segmentation model to crop the body from the video, and (b) a pose estimation model to obtain the skeletal keypoints. Our frontend is built with flutter. The activity data is retrieved from a Garmin smart watch via Garmin Connect and Fitrockr.

Challenges we ran into

Cropping the video and recognizing the joints and skeletal keypoints correctly, was a challenge.
Getting reasonable recommendations based on the available data.

Accomplishments that we're proud of

Built something with impact we would definitely use by ourselves
Finished before deadline (hopefully ;))

What we learned

Gained further experience in flutter, video segmentation and pose detection
Time management
Pitching & Storyline
Correct execution of squats

What's next for Line - Stay Injury Free

As we see a direct impact on our personal lives, we are strongly motivated to bring Line to the next level. As next steps we want to improve our data analysis to help more people to stay injury free based on predictive maintenance.