Crowd Counting

Product information

Figuring our how many people are in a scene is a complex task, stop putting a person in the field to manually count the different people passing, instead upload your scene and let us do the rest!

Image Cover Difference
Image Cover Original

How to use the Crowd Counting

Learn how to get started in just a few simple clicks!

Our policies

  • 1. Upload Your Image

    Start by selecting your file, which you can upload by File or URL.

  • 2. Run Crowd Counting

    Once you have selected and uploaded your file, click "Create" to run the Crowd Counting tool.

  • 3. Download Result

    As soon as the Crowd Counting finishes, you can download the result.


How does Crowd Counting work?

Crowd Counting is the process of counting individual people within a complex scene that presents issues for normal person identification algorithms. Crowd Counting differs in this way as it looks as t the probabilities of people standing next to each other.

Crowd Counting will take in a picture and detect the heads of the people within this image. Afterwards, a separate neural network is utilized the predicts the likelyhood of additional people being next to each other.

The process of Image Matting can be described in a few simple steps:

  1. PreProcess: We resize the image to an integer size (% 32 possibility).
  2. Infer: We run the image through the neural network that was trained for this task.
  3. PostProcess: After the result is returned, we filter out the estimation and location of the different detected people based on a threshold.

We’re here to help

As we simplify this entire process, by creating the model for you. Training state-of-the art models by their latest paper performances, while providing the transparancy you as a business require for what is happening behind the scenes.

Workcation takes away the pain in using state-of-the art AI models. Test and integrate AI into your application in just a few clicks.

Xavier Geerinck

CEO at


Find some examples below of what you can achieve!

  • Image Masked
    Image Original
  • Image Masked
    Image Original
  • Image Masked
    Image Original

Ready to dive in?