How do I find a parking space in 5 seconds

The large size of your driveway does not guarantee that you will find a parking space quickly and easily, and more often than not, you can get depressed when you pull into a parking lot because you’ve driven around the lot, wasted time, and ended up not finding a space. How you can add a little more certainty to your life, I’ll describe below.

Many of us, like me, live in classic man-houses.

Someone is lucky and they have underground parking, where you can buy a parking space, by the way, not a cheap pleasure, but in most cases, they roll up the territory with asphalt with a limited number of seats, disproportionate to the number of tenants – car owners.

Usually, I come home in the evening and see people scouring the parking lot for space or sitting on their cell phones waiting for someone to leave. I do that myself sometimes.

Most of the time I call my wife and ask her to look out the window and tell me where there are spaces. But it is not always stable, so I decided to replace it with something more automated and working even in the late hours.

So, what we need: is a camera(s) preferably over the parking lot, a service with a user-friendly UI, and a back that will look for available seats in the video.

read also : Overview of data backup and recovery systems as IBM Spectrum Protect

Camera(s).
Usually, in new buildings, there is a video surveillance system, so the video streams from the cameras can be taken from there. What disadvantages I see:

  • You have to negotiate with the HOA to get “legal” access.
  • A small viewing area.
  • As a rule, the cameras do not cover the whole parking lot.

An example of the location of cameras in my housing estate
Example of the location of cameras in my housing estate
It is best to place a camera on a roof or in an apartment with a clear view of the parking lot. In my case the second option is the best – I live quite high up.

In general, the camera can be anything, but there are a few tips that I have learned when choosing a camera:

  • The higher the resolution, the better the object detection will work, so I suggest considering cameras from 2MP.
  • There should be a large viewing angle of >90 degrees – this is needed to solve the problem with fewer cameras. Choose cameras with 2.5 – 2.8 mm focal length and > 90-degree viewing angle.
  • Day/night mode will be a big plus.
  • An optional parameter – IR illumination, I’d even say it’s questionable, in practice is needed only if the camera is placed outdoors, and given the distance and weather conditions, it can only make things worse.

If you have a camera it does not mean that everything is ready. You need to install it, configure it, and somehow route video from the camera to your server, which will analyze the video for empty parking spaces.

I took measurements, calculated the installation location, and used precision shoe boxes to install the camera according to all the techniques.

Setting up the camera involves turning off the IR light so it does not reflect off the window in the evening and turning on the automatic day/night mode. Day/night mode, roughly speaking, is the camera’s ability to switch from normal, color mode to black and white. As a result, the parking lot is also perfectly visible in the evening.

There are several ways to route video from the camera to your server:

  • Place a server at home (on the same network as the camera). Works as a temporary solution.
  • Through proxy software located in the same network and output/proxy video to our server. The disadvantage of this is that this software must be somewhere “rotate”, ie it needs a constantly running machine. The same as the server at home.
  • Setting up DDNS on the camera. The same proxy services are available for the camera, as a rule, it is impossible to change the list of such services on the cameras, and foreign services are not always convenient to use.
  • Buying a static IP address from your Internet provider. This solution has both pluses and minuses, but to add your camera on a remote server, it is enough to make two clicks in your cabinet and set up port forwarding.
  • Set up a VPN on your router to the server.
  • Through integration with the cloud service of the camera. Most cameras now have a service that allows you to view video from anywhere in the world without much effort, but sometimes it is impossible to integrate with such a service.
  • etc.

For me, the easiest, fastest, and the most convenient option was to buy static from the provider.

Service with a user-friendly UI

Interacting with the service should not distract from the road when approaching your house. Also, the service should be publicly available, understandable, and easy to connect and configure.

A simple, quick, and convenient solution are to make a bot. There are a million articles on how to make your bot, but I won’t describe them here.

You choose a bot in messenger, click “Find Parking,” and there you have a picture of free parking spaces. It doesn’t get any easier than that.

A backend that searches for free parking spaces.

There’s a cool article translation (looking for a free parking space with Python) on how you can implement a car search with Mask R-CNN. The article uses an automated approach to mark up a parking space. In the case of adjacent parking, it is better to mark it up in advance. This will save us from false positives, if cars are blocked due to lack of space, special services cars stop, parking on the lawns/snowdrifts, etc.

An alternative option is to use an off-the-shelf solution with easy setup and API. Macroscope company, where I work, deals with intelligent video surveillance, we have a function of counting objects, and we use YOLO neural network by the way. For me, this option is the most acceptable because I have a finished service that provides car recognition analytics with GUI configuration of parking zones and API.

For clarity, a complete flowchart of the service

The user clicks “Find Parking” in the Telegram bot. The service bot requests the coordinates of free spaces from the video service. If a spot is found, it requests a picture from the camera. The bot service draws the obtained zones on the picture and returns it to the user.

The bot can add parking. This is done to separate users. For example, you can add parking in front of work or another LCD.

If you want to try it, then write in a personal message and I can connect your house to the service without problems.

What will the result

My residential complex has over 900 apartments and about 150 parking spaces. On the first day, the bot added 60 + people. I did not expect it at all, but without any PR the number of users grew to 100 within the first month. When I re-posted the link to the bot, the number of users increased to 144. Also unexpected and nice is that on average 25 people per day are using the bot.

Judging by the feedback, the main feature is a highly placed camera with the ability to quickly get a picture from it through the bot.

The main problems with setting up the service:

  • Setting up the camera is not quick/easy.
  • Lack of accurate detection of free places (can be solved by training neurons).
  • In summer everyone has vacations and summer cottages, so there are more free parking spaces, so the need for the service decreases dramatically.
  • My wife is not happy with the installation on the windowsill.
  • Installing a camera on the facade should be coordinated with the MC and the tenants. Without this, in my case, in the morning hours, it is impossible to use the service because of the light.

Leave a Reply

Your email address will not be published.