This is my digital posture monitoring app that helps you sit properly and maintain good posture while working at your computer
it uses your webcam to get the camera feed and then uses Resnet50 model from tensorflow to analyse your body posture. then it uses those pointers to calculate the distance between your shoulders and head it also takes in consideration the distance between your shoulder and the orientation of both the shoulders and head then it gives you a score after evaluating these pointers
kashsuks
Check their projects out: Yapper - A slack bot for my channel, Go Games, Buckrise, Neohack - YSWS Site, CPVim, CPCli, CFCli, Domain Expansion (Personal Site)
Once you ship this you can't edit the description of the project, but you'll be able to add more devlogs and re-ship it as you add new features!
Completed the whole project, added minor suggestions as suggested by the community for some enhancement or lagging related issue, the Readme is also completed now!!
my MACBOOK is One Step Closer if you press that vote button 😆
Reworked on the slider and now you can experience at both very fast frame process rate 0.1 or very slow rate 99.9 also did some minor changes in the model fine tuning
PLEASE WATCH THE VIDEO it has live demo ( it took 4 attemps because my system was crashing)
Added export button, improved the frame rate counts, added performace tracking, improved the point system. Now everything works smooth
Changed the logic behind working of the model and points system, added slider to change the rate at which frames are procesed so that if you dont have too much battery you can switch to a higher rate and fixed the buttons to be responsive
Added achievements feature, currently added 6 achivements for consistency and motivation, tested 2 of them need to test the others also and need to fix some js for this
So it was still lagging a lot and then I noticed that its because of those pointers still I modified the whole code to make it more optimized but I also added a optional button to change that and toggle between showing the pointers and skeleton or not
Improved the settings functionality by adding detection ,notification,visual and performance settings
Made the good and bad timing working previously it was lagging a lot, changed the base value to 40 and the good bad to 60 so now its more accurate
added the break exercises and added tracking so that we can track if the use is actually doing them or not
added the ResNet50 model after testing with MobileNetV1 because resnet performs better and then added the pointer for the full body and the score also works pretty well now
IDK why but electron js is fucking up with camera access so I need tried online to look for resources but I found nothing so now I need to switch the platform maybe to web interface
Made the whole UI in electron js and now need to implement the camera functionality to get the data