PD

3DToF People Detection and Tracking User Guide

Overview

The 3DToF People Detection and Tracking app allows you to simulate and analyze the movement of people and objects detected and tracked by ADI’s Time-of-Flight technology.

You can test the app’s functionality by using the sample dataset mode. Select one camera module from the predefined list containing ADI’s 3D Time of Flight technology.

Upload your own files and get the results by customizing the following variables:

  • Settings, such as Background Subtraction Learning Rate, 3D Points to Object Grouping, Depth Scale, FPS Limit
  • Region of interest
  • Camera Lens Parameters, such as Intrinsic Coefficients and Distortion Coefficients

Analyze frame by frame the results provided in various forms: point cloud, IR image, and 2D detection.


Getting Started

  • Access the vLab page and navigate to Apps.
  • Select the 3DToF People Detection and Tracking app.
  • Click Open App.

Once a new project is created, you can see the Configuration field to your left and the simulation panel on your right.

/assets/user-guides/people-detection-overview.png

Figure 1: Details for this figure go here


Configure

1. Choose the Dataset Mode

You can choose between two different dataset modes for the simulation:

1.1. Sample Dataset

If you don’t have any images and you want to test the app’s functionality, you can use the sample dataset. Select one camera module from the predefined list containing ADI’s 3D Time of Flight technology.

  • AD-96TOF1-EBZ
  • AD-FXTOF1-EBZ
  • AD-3D Smart Cam1-PRZ

Then select a sample dataset from the dropdown menu. Each sample dataset has its own specifications that will be displayed below for your review.

1.2. Upload Dataset

If you want to use your own images from your local machine, choose the upload dataset mode. You’ll need a pre-recorded dataset when choosing this mode. When preparing your images, please keep in mind that the app supports only BIN files with a maximum size of 2 GB.

After uploading your images, add the camera module name in the corresponding field. If you’re using ADI technology when recording your images, this field will be updated automatically.


/assets/user-guides/people-detection-dataset-mode.png

Figure 2: Dataset Mode

2. Edit the Optional Fields

Once you have chosen the dataset mode, you can edit the optional fields before running the simulation. However, none of the variables are mandatory; therefore, you can run the simulation with the default values.


2.1 Settings


To edit the values, manually enter the desired value or click “+”(plus) to increase it and “-”(minus) to decrease it. Only numerical values are accepted; alphabetical and special characters are not accepted, except “-“ (minus).

  • Background Subtraction Learning Rate – accepted values are between -1 and 1.
  • 3D Points to Object Grouping – accepted values are between 0 and 4.
  • Depth Scale – accepted values are between 0.1 and 1.
  • FPS Limit – accepted values are between 0 and 30.0.

2.2. Region of Interest


To edit the values, manually enter the desired value or click “+”(plus) to increase it and “-”(minus) to decrease it. Only numerical values are accepted; alphabetical and special characters are not accepted, except “-“ (minus).


  • 3D Space – accepted values are between -10 and 10. For 3D space, you can configure X, Y and Z minimum and maximum values.
  • 2D Space – accepted values are between -10 and 10. For 2D space, you can configure X and Y minimum and maximum values.

2.3. Camera Lens Parameters


The camera lens parameters fields represent the correction coefficients for the lens. To edit them, select or unselect the correction coefficients and manually enter the desired value. Only numerical values are accepted; the minimum value is 0 and there’s no maximum value. The numbers can have up to 3 decimal points (e.g.: 1.123). The following correction coefficients are available for editing:


  • Intrinsic Coefficients: Fx, Cx, Fy, and Cy.
  • Distortion Coefficients: K1, K2, P1, P2, and K3.

/assets/user-guides/people-detection-camera-lents.png

Figure 3: Configuration Panel




Run the Simulation

After inputting (and double-checking) the values in your configuration, click Run.


Within a few seconds, the 3DToF People Detection and Tracking app will analyze the dataset you offered, and it will generate the results.




Analyze Results

Each results tab will present with the following data for you to analyze frame by frame:


Comparison tab – displays all the generated images at the same time.

Point Cloud tab – displays a 3D representation of the simulated results.

IR Image tab – displays an IR image of the simulated results.

2D Detection tab – displays an IR image of the simulated results, with a bounding box on the detected people.


/assets/user-guides/image.png

Figure 4: Results




Additional Feature

  • To pause and resume the simulation results, click the pause/play icon located under the simulated results.
  • To move from one frame to another, click the previous/next icons located under the simulated results.
  • Choose from various speed options: 0.25x, 0.5x, 0.75x, Normal, 1.25x, 1.5x, 1.75x, 2x.