Traffic Crowd Monitoring and Analysis
Traffic crowd analysis is one of the modern research problem. Nowadays traffic congestion and traffic signal violation are serious issues in many metropolitan cities. Such problem result into collision of vehicles at traffic signals. An efficient traffic management technique is needed to reduce wait
2025-06-28 16:29:51 - Adil Khan
Traffic Crowd Monitoring and Analysis
Project Area of Specialization Artificial IntelligenceProject SummaryTraffic crowd analysis is one of the modern research problem. Nowadays traffic congestion and traffic signal violation are serious issues in many metropolitan cities. Such problem result into collision of vehicles at traffic signals. An efficient traffic management technique is needed to reduce waiting and travelling times, save fuel and money. Using video stream from the existing cctv cameras installed over the roads, we want to analyse the traffic crowd by:
- Detecting vehicles and counting them,
- Measuring crowd flow,
- Detecting obstacle.
We can use this information to address many problems, including, but not limited to:
- Detecting traffic congestion hotspots in an area,
- Detecting obstacle/accident hotspots,
- Smart Traffic Light system.
We will be researching the state of the art work on these problems, try running these algorithms using pre-trained models on our local data. Then we'll train the model on our local traffic data and fine-tune it to get increased accuracy, for which we wil be needing GPUs. For getting the local traffic data, we'll be getting in touch with Punjab Safe City Authority, try to benefit from their experience and brainstorm on the problems they are currently facing in monitoring the city's traffic.
Project Objectives- Detecting traffic congestion hotspots in an area:
- Using the measured statistics to check which areas are mostly affected by traffic congestion at what timings. This will help the traffic police to create diversions to ensure smooth traffic flow.
- Detecting obstacle/accident hotspots
- System will be able detect obstacles on road such as dead animal and generate warning for the authorities to take preventive actions.
- Smart Traffic Light system.
- System will be able to control traffic signals using the measured statistics. Such as it will count the number of cars in a signal to set the time of traffic signals accordingly. This will also reduce traffic congestion and will also help reduce air pollution.
- Collecting data from various sources such as (Punjab Safe City Authority) and data available on internet.
- Studying state of the art research papers on traffic density analysis and reimplementing their algorithms on our dataset.
- Training the state-of-the-art traffic density analysis algorithms on our dataset.
- Tweaking the state-of-the-art algorithms to address our objectives.
- Using the measured statistics to implement machine learning models to predict traffic congestion.
- Using the measure statistics to implement the smart traffic light system.
- If necessary, make our own algorithm to address our objectives.
- Automating the Traffic lights will
- Make the movement over the roads, time efficient.
- Reduce the traffic pollution.
- Reduce the surveillance required by the traffic wardens.
- Detecting traffic congestion hotspots will guide the government decisions on improving the road infrastructure and to create diversions.
- Detecting obstacle will help in guiding the citizens in choosing the best route for their destination.
- A sophisticated software system that will take live cctv stream and display the video with calculated statistics.
- A map of an area/town/city with statistics, including, but not limited to: traffic hotspots, accident/obstacle hotspots, current state of traffic count and flow.
- Automated system for controlling traffic signals
| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 30250 | |||
| Google Colab Pro License | Equipment | 1 | 8250 | 8250 |
| Travelling cost | Miscellaneous | 2 | 1500 | 3000 |
| CCTV Camera | Equipment | 3 | 5000 | 15000 |
| Data Annotation | Miscellaneous | 2000 | 2 | 4000 |