The problem parkIn solves

Currently unauthorized parking detection systems mostly rely on the manual way of verification and the generation of parking tickets. There is no centralized way of keeping track of fines and vehicles in the database. Hence, a modern solution is urgently required to address the shortcomings of current systems.

Our initiative, parkIn aims to combine video
surveillance and efficient parking using CNN
(Convolutional Neural Network) through YOLOv4-tiny.
Using the concept of distributed systems, we aim to
channel the live feed from the CCTV cameras to a
processing server where our algorithm will detect if the
vehicle is parked correctly or not in the right area.
The processing server will relay feedback and will alert
the authorities and the owner using SMS/email.
A web-app is designed where the vehicle owners can
check their fines and dues for wrong/unauthorised
parking.

Challenges we ran into

Tensorflow and ML Model issues : we got over by reading the docs and watching examples.