Project: Deep Learning methods for attendance/authroisation automation

[hdidwania]

I have been thinking that we can give the college some automation by using Deep Learning, and save the students from unnecessary and stupid procedures (like signing on register while leaving campus, etc). Also maybe we can think of a better attendance system (e.g. camera to scan ID cards of students while entering class, maybe?). There is a lot of room of automation in the institute, which can yield powerful results. Also some physical automation task can be achieved using AI + Microcontrollers/Microprocessors.

If the AI community of college feels good, let's pick up one small project and try to build it over this semester. Feel free to suggest ideas.

[ayush1999]

I don't really have a good knowledge about MPMC and electronics stuff, but how do you propose to use Deep Learning to automate events like signing on the register while leaving campus? Also, in full support of implementing scanner for IDs inside classrooms, atleast it'd make the process of proxy simpler.

[hdidwania]

how do you propose to use Deep Learning to automate events like signing on the register while leaving campus

I thought that we may implement a face recognition system which can use cameras installed at main gate to automatically detect and recognise students moving in and out of campus and hence automatically putting up the names in database, hence not requiring anyone to go in and sign that stupid stuff up. But this again, has lot of challenges, and maybe used by college authorities against us students. Need to think over this, Another automation idea was given by the robotics people last year, was automating the main gates, operating it without human force ie guards won't need to go and open it. We can use cameras to detect a vehicle is approaching, to open the bigger gates, or a pedestrian of a 2 wheeler is approaching, to open to smaller one. Again a lot of challenges to such a problem statement, but can be explored.

implementing scanner for IDs inside classrooms, atleast it'd make the process of proxy simpler.

If carefully planned and executed, it would help us students a lot. We need to make sure though that the authorities/faculties never get suspicious about it. One theory I have is if the stuff is automated, they won't bother checking its accuracy and cross checking the results, but this may be a faulty assumption which will put us in trouble if cross checked.

[saiamrit]

For the attendance system may be we can keep it simply to a web app that can just use web cam to capture face and recognise face and automatically update attendance database.we can try out with a openCV thing or else write a deep net.

[ankitjha97]

@hdidwania , it's quite simpler one.There a library called dlib which you can check. Main think is to integrate it.

[IamRaviTejaG-zz]

----- SNIP ----- (like signing on register while leaving campus, etc). ----- SNIP -----

How about placing a QR code at the exit point, and make an app where the user will stay logged in using his Mandaar UMS credentials, and once he scans the QR, he is granted exit permission and the permission stays alive for 6 hours (assuming you leave at 5.30 PM & return at 9.30 PM, even then you have a cushion of 2 hours extra), and scanning the same QR grants the entry?!

Just my opinion. I believe the entry-exit thing has a simple approach. But regarding the attendance in class, yes, I believe ML can be put to use. :)

[arijeetsat]

How about placing a QR code at the exit point, and make an app where the user will stay logged in using his Mandaar UMS credentials, and once he scans the QR, he is granted exit permission and the permission stays alive for 6 hours (assuming you leave at 5.30 PM & return at 9.30 PM, even then you have a cushion of 2 hours extra), and scanning the same QR grants the entry?!

This same thing one of our senior has shown to Director sir, he said "there is no need, we also don't want to bind you ", if someone will implement face recognition system, then that he can accept.