Fraud detection with Machine Learning and Computer Vision
$30-250 USD
Suljettu
Julkaistu noin 6 vuotta sitten
$30-250 USD
Maksettu toimituksen yhteydessä
Into a ice cream shop we have a security camera covering almost all the room. On the other hand, we have a cashier system where all sales are recorded.
An employee dispatch the ice cream and record the sales on the cashier as customers come to shop. We have real time access to both, security camera and cashier records.
This project is about search a way to detect ice cream dispatch without record on cashier. We'd like a executable file (you can pick the language) doing analytic in real time and output a line to stdout when a threshold is reached.
Hi. My name is Uzair.I am in final semester of my masters in Electrical Engineering. I am doing my thesis in biomedical signal processing and Machine learning.
I have more than 3 years of experience in MATLAB specially in Machine learning,Signal processing and image processing.
I think I can handle your project with quality work. Lets discuss.
Regards
Dear, I am Emanuele Martini a biomedical engineer who works as bioimage analyst for a research campus in Milano.
I work daily with matlab and python and other tool to analyse images. I think I can help you to debug your code.
A part of some freelancer projects where I used computer vision to identify objects.
Please contact me if interested in my collaboration, have a nice day Emanuele
I am a specialist and sure of what I can do and convey. I can deal with Arduino, Python, Java, c+ Matlab, among others. I am requesting an opportunity to work with you on your undertaking.
Hello,
I think I could bring you a lot to realize the project.
The idea for solving the problem is to look for the timestamps where the recordings of the camera without the use of the cashier system are close to the recordings of the camera that are associated with a use of the cashier system.
Because we have 3 scenarios that the camera can film:
1) No sale of ice + no use of cash system
2) Sale of ice + use of cash system
3) Sale of ice cream + no use of cash register system => Fraud
In this case, if we consider just camera records, the case most similar to the 3rd case is the 2nd.
So we can use some sort of Euclidean distance between records where there was no use of the cash system (case 1 and 3) and records where there was a sale plus the use of the cash system (using just camera records).
We have to define a numerical threshold to say whether it is fraud or not.
Thank you for your attention.
I hope that my idea will satisfy you. And I thank you for your trust.
Have a good day.