search

ruxailab / web-eye-tracker-front

MIT License
6 stars 22 forks source link

Feature: Create a realtime prediction screen after callibration process. #1

Open marcgc21 opened 7 months ago

marcgc21 commented 7 months ago

(edited by @KarinePistili )

Create a realtime prediction screen after callibration process.

First follow this flow: https://github.com/ruxailab/web-eye-tracker-front/issues/35#issuecomment-1979195869 to create a calibration. Then you should grab the model and use it to perform real time predictions. This task requires you to create a new page after the calibration and try to use the model provided by the calibration to create a real time prediction environment.

akoolarni commented 4 months ago

can i get to more about this issue please?

hvini commented 3 months ago

Hello @akoolarni. of course.

First follow this flow: https://github.com/ruxailab/web-eye-tracker-front/issues/35#issuecomment-1979195869 to create a calibration. Then you should grab the model and use it to perform real time predictions. This task requires you to create a new page after the calibration and try to use the model provided by the calibration to create a real time prediction environment.

hvini commented 3 months ago

The branch feat/real-time has an initial implementation of this feature you can take a look into it for inspiration.

You can use the code on the api project to generate the model used on js: https://github.com/ruxailab/web-eye-tracker The api branch is feat/realtime

akoolarni commented 3 months ago

Thanks a lot @hvini for the help

akoolarni commented 3 months ago

Will work on this

xr-dev-saurabh commented 3 months ago

Hi @hvini. If this issue is still open, can you assign me this issue.

SujashB commented 2 months ago

Hi @hvini! I submitted a proposal for Google Summer of Code to try to fix this issue. Can you please assign it to me so that I can start working on it?

KarinePistili commented 2 months ago

Hello, as this issue is being inactive for some time, I will restart the interest of assignment. I've update the description to be more clear as well.

The idea behind this issue is to:

Create a realtime prediction screen after callibration process.

First follow this flow: https://github.com/ruxailab/web-eye-tracker-front/issues/35#issuecomment-1979195869 to create a calibration. Then you should grab the model and use it to perform real time predictions. This task requires you to create a new page after the calibration and try to use the model provided by the calibration to create a real time prediction environment.

KarinePistili commented 2 months ago

Please comment here if you want to work on this :point_down: :point_down: :point_down:

SujashB commented 2 months ago

interested!

KarinePistili commented 2 months ago

assigned @SujashB ! If you have any question feel free to ask them here

SujashB commented 2 months ago

To do the real-time prediction, what techniques do you suggest I use? I was thinking that I use KNN classification and Neural Networks with 2 hidden layers to do prediction.

KarinePistili commented 2 months ago

Hi @SujashB , this particular issue is about using what we already have on the project. So it is to see how our linear regression model would behave with someone using it.

You need to add a screen after the callibration process that uses the data of this model to create the real time page.