International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 12 Issue: 11 | Nov 2025
p-ISSN: 2395-0072
www.irjet.net
A Computer Vision–Enabled Smart Parking Framework – A Review Prabhakar Khandait1, Devanshu Awalekar2, Prashik Tapakire3, Isha Upare4, Prathmesh Nagpure5, Raj Pardhi6 1Head of Dept., Dept of Electronics and Telecommunication, KDK College of Engineering, Maharashtra, India
23456UG student, Dept of Electronics and Telecommunication, KDK College of Engineering, Maharashtra, India
------------------------------------------------------------------------***-------------------------------------------------------------------------
Abstract - The inefficiency of parking in the cities is one of
job spotting vehicles and empty parking spaces across various conditions, making results steadier. Still, these smart systems ask for heavy computing power, which might block their use on low-end gadgets, so you’ve got to weigh accuracy against performance. Fresh forecasting methods say, reinforcement learning or predictive algorithms - help look ahead by guessing how things will be used and prepping resources early. But they need steady, clear data to work well, falling apart fast if usage spikes out of nowhere.
the greatest problems in the contemporary cities that result in traffic congestion, fuel wastage, and frustration by the drivers. This paper includes the design and implementation of an AIbased Smart Parking Framework that will utilize real-time slot detection and availability analysis to optimize the use of parking space and allow a better user experience. The proposed system incorporates computer vision algorithms, IoT-based sensing, and edge computing to detect vacancy slots correctly and predict future availability according to the occupancy patterns. A user-focused interface will have realtime updates that will decrease the time taken to conduct a search, decongesting traffic. The efficacy of the framework is experimentally validated using the improvement of the detection accuracy, reduction of latency, and scalability of the implementation in the environment of the smart city. This paper illustrates that artificial intelligence and IoT convergence can be applied to convert the conventional parking system into an intelligent and data-driven approach to sustainable urban mobility.
Despite improvements, today’s smart parking tech keeps running into problems. When just one part crashes, there's no fallback option available. Cloud-run models tend to lag and eat up lots of bandwidth. Merging different systems remains tricky, plus city-wide scaling often falls short. That means we need a unified setup - able to blend multiple sources, make quick calls using on-site computing, but also predict changes to stay adaptable. This research fills current holes by launching a clever parking setup powered by AI, mixing camera tech with connected devices for steady, breakdown-proof sensing. But rather than lean only on distant servers, it runs small AI bits right inside gadgets - spurring fast reactions and space to expand. Predictive features bring some adaptability; still, the main leap lies in practical edge computing paired with fused sensor-and-camera data. So, this linked strategy lifts reliability, cuts lag, yet keeps city-wide deployment simple.
Key Words: Smart Parking, Artificial Intelligence & Machine Learning, Internet of Things (IoT), Real-Time Slot Detection, Availability Analysis
1.INTRODUCTION City traffic's becoming a real headache these days. With more folks moving in and vehicles piling up, parking setups can't keep up. When spots aren't handled well, streets clog up - fuel gets burned for nothing, fumes increase, drivers get frustrated. It messes with everyday routines - not only that, it drags down communities, wallets, and nature alike. Fixing how we manage parking plays a big role if we want cleaner, sharper urban areas.
The study makes a difference in two ways. First, it boosts academic talks by mixing solid parts from old techniques into a single setup - fixing ongoing problems like lag in detection and scaling up systems. Second, it gives a practical framework built for real city settings, particularly intelligent parking spots, linking ideas to real-world action. Instead of aiming for small tweaks, the method focuses on building stronger foundations. That way, urban parking gets more dependable, less harmful to nature, while staying easier to use
The response to these problems included tech for smarter parking. At first, systems tried spotting open spots with cameras; but they usually failed when light changed or something got in the way. Rather than just using video, updated models added sensors along with internetconnected tools for better results. Even though those did fine when things stayed predictable, fresh challenges popped up like lag, high costs to install, and constant maintenance.
The rest of this paper unfolds like this. The next section gives a close look at today's smart parking options, but also checks their strengths alongside weaknesses. After that comes a part revealing gaps in earlier research as well as explaining why the new setup looks the way it does..
Right now, tasks are handled nearer to where data comes from - thanks to edge or hybrid setups. This slashes lag, uses less network space, yet makes reactions quicker and more dependable. Meanwhile, advanced neural nets do a better
© 2025, IRJET
|
Impact Factor value: 8.315
|
ISO 9001:2008 Certified Journal
|
Page 919