-
-
basic ui without video
-
ui with image an posture ratings
Inspiration
In a world dominated by computers everywhere we go, neck posture and eye strain, has become an issue that affects everyone. Living on a college campus, it is not abnormal to see people hunched over their computer screens, often in stress trying to learn the content for a class. This stress is only multiplied by the pain caused by their sustained poor neck posture, as well as a lack of break from the screens, which causes eye strain. Thats why we created this app, in order to allow students and anyone using a computer to keep their neck free of stress, and keep their eyes hydrated.
What it does
Tenjin uses gemini, to track your neck posture, your proximity to screen, your eyelids, and your pupils, in order to alert the users when one of these metrics is unoptimized for a risky period of time. The app has an audio cue that will let the user know, and the user can see exactly what is wrong with their current state, and is shown how to fix it. The app can also tell when the user is distracted when the eyes deviate from the screen for a long period of time. At the end the app gives the user feedback on all their metrics, and how those metrics can be changed in the long term.
How we built it
The code was built using google ai studios, and gemini pro 3, after tons of trials where we changed the logic of how the app would work. We also used Claude.ai in order to help us fix the code and in the end after 7 different prototypes, and many different lines of logic, we finally were able to create a system that was fully optimized, and able to run faster with the use of AI.
Challenges we ran into
Our biggest challenge was for sure the logic of how the app would work. The original sensor used by google ai studios, is very flawed, especially in the latency which cannot be changed. To combat this the team worked together and we found a very viable solution, which was allowing Gemini to check the frame every few seconds, and analyze the users posture, to give the most accurate feedback. The ai was much faster than the sensor we used originally, because the ai was able to observe the ratio between our shoulders and neck in a much more efficient way, which just a regular code would struggle with. Another big struggle we faced, was that after adding changes to the app over long periods of time, all versions of the app would just cease to work, which meant we had to create a new prototype, which created the perfect balance between functionality, but not overwhelming the Ai coder. To do this we learned something from every prototype, which was implemented into the next. In the end we got that perfect blend, which allowed the app to run smoothly.
Accomplishments that we're proud of
The biggest accomplishment as a team, was actually making the app function with the ai after we found that perfect blend. This is because, it required all of us to work tirelessly, from coming up with the idea to have the ai analyze the frame rather than use a sensor, to having each of us brainstorm ideas on how to change each prototype in order to reach that perfect blend. I am very proud of my team members who put in the work, even after countless hours with plateaus in progress, we persevered and were able to optimize the app to work very well.
What we learned
We think the biggest lesson we learned, was the importance of brainstorming different paths of logic in order to make the code work seamlessly. Anyone can just ask ai to make it a code, but not everyone can create that perfect path of logic, which allows the app to work seamlessly. There are so many ways to go about this app in terms of logic, but I truly believe that the one that we as a team thought of, was perfect for the app.
What's next for Tenjin's Vision
First of all we will allow the users to create an account and add Tenjin's Vision to a public domain so that anyone can use it. Using the account Tenjin's vision can monitor the users neck posture, and eye strain over many days, and give the user feedback. We also wish to add more features, such as the ability of the app to allow the user to notified when they are biting there nails, or participating in similar bad habits. We also want to work to release this app on app stores, on android and apple, and possibly scale it with a paid model.
