search

derm1ch1 / spARkle

Software code for the spARkle appliance
0 stars 1 forks source link

Implementation #1

Open PraveenGit3 opened 3 years ago

PraveenGit3 commented 3 years ago

Hi , I m intersted in the implementation and could you able to share your notebook for seeing how its performs on other images too? I installed all dependencies but still no luck Unfortunately I m unable to run you detections could you add readme file and .ipynb of your implementation .

derm1ch1 commented 3 years ago

Hey @PraveenGit3, thanks for your interest in my project! I'm sorry for the inconvenience - this project is still under development.

I do not have a Jupyter Notebook installation for this project atm. - but I am currently writing a detailed installation guide. If you are interested I can send it to you and after we fixed your issue I will document the solution here. :-)

Looking forward hearing from you! Best regards

PraveenGit3 commented 3 years ago

thanks for the response . could you tell me how did you approach solution for web application? i could undertsnad architecture behind Prediction part ie Detection of parked and vacant spaces but how did you able to pass real time infernce results in app ? I means if model detected 10 vacant and 5 occupied slots ,same is reflected in app ? Did you streamed data to web server and run the model there itself and saved to webserver and diplayed the boxes in website? I would love to connect with you on this or some explanation

derm1ch1 commented 3 years ago

Hey, sorry for the late response, I wanted to attach you the files for the webapp first but I need some more time since I'm currently on a business trip and have no access to the files.

But just to give you an update: I understand your question. But just for you to keep in mind: This software was developed as part of my bachelor studies. Therefore I "just" save the analysed results every 20 seconds in a flat-file (.json) on my hosting server. I then use these data in my webapp and refresh every minute to have the current data.

So you are correct, the current parking space situation is relfected in real time in the app. The webapp nevertheless does not run the model or something, for ressource reasons it's only displaying final results.

I will add a www-folder with all parts of the webapp into this repo. I wrote it completely myself without any higher framework and designed a custom parking space UI you can modify for your needs. It's also possible to exchange the flat-file with a database - this is what I am testing currently: Write the analysed data into a SQL database every few seconds with python on the edge device --> then make a request in the webapp to get these data and display them in the UI.

Have a good day!