Published on August 19th, 2018


Can we eliminate traffic congestion with the help of Computer Vision?

Writing out my thoughts on eradicating traffic congestion in highly populated cities.

4 minutes read

Whether it's being implemented or not, I have been thinking on how we could use Computer Vision to solve traffic congestion.

Since, I live in the United Arab Emirates, I have always observed that people who commute from Sharjah to Dubai and vice-versa face a lot of traffic jams despite all the new roads and toll-gates (yes, I don't seem to understand how does that solve the problem).

Traffic in Al Ittihad Highway.

Well, the problem is not only faced in this country but many countries such as China, Indonesia and so on.

What are the causes of traffic congestion?

Anyways, I jotted down some facts to consider what causes traffic congestion in the first place:

I'm sure that there could be more but these are the facts that I can come up with for now.

How can Computer Vision solve this problem?

Computer Vision is a field that intersects with multiple areas of technologies in which it aims to develop an understanding of objects by extracting information from various digital media sources like images and videos and automate those tasks that a normal human would do in their daily lives.

Visualization of Computer Vision.

There are various types of problems that Computer Vision algorithms are able to solve but not limited to:

Now, it's not only about implementing these CNN-based (Convolutional Neural Network) algorithms but you also need hardware to be able to compute and process data. 

How would this work?

There are two scenarios that I had thought while writing this article and I hope that I'm able to translate my thoughts into accurate examples.

Let's pretend we have four car drivers: Alex, Bob, Charlie and Dylan.

Speed-Distance equilibrium

Alex, Bob and Charlie are driving on the same lane. Alex hits the brake slowly to shift to another lane, the sensors of Bob's car detects a change in speed in Alex's car, Bob's car adjusts it's speed to match Alex's current speed based on the variables of distance and travel time, Charlie's car adapts the changes of Bob's car and thus, it's a chain reaction.

Shifting from one lane to another

Alex is driving in Lane A and Dylan is driving in Lane B. Alex wants to shift to Lane B, so he switches on the indicator and Dylan's car sensors identify that Alex's car wants to change lanes. So Dylan's car slows down and Alex is able to shift lanes with ease. I thought of it to be some sort of a "elastic" effect when this occurs.

Well, you might argue that some cars have a system called "Cruise Control" but here's the part that I'm trying to pitch, as I had mentioned above, I just wanted to integrate sensors to the front and rear sides of a vehicle, which means that these sensors can be integrated in almost any vehicle.

How is this going to be helpful?

For starters, traffic congestion will gradually reduce, if not, be eliminated but there are other beneficial factors to it, such as:

However, if the sensors fail to work, the car driver will still be safe because it's surrounded by other cars that have the sensors. This made me think of another question, does that mean do all cars require sensors or only a few? I find it quite intriguing.


Although, these sensors might require a vehicle to have some intelligent capabilities, it may not require the type of network found in an autonomous vehicle.

The idea of placing sensors in the front and rear of a vehicle can optimize the flow of traffic and thus, it might help eliminate traffic congestion.

Hope you liked reading this article!