Skip to content

Plumerai People Detection demo on Arm/x86

This document describes the use of the people detection demo: a ready-to-run binary to evaluate the Plumerai People Detection library by running it live with a camera. To get access to this demo, you can fill in our contact form and ask about the ESP32 demo. The Plumerai People Detection demo package consists of:

  • The plumerai_demo compiled binary executable(s).
  • The libplumerai_gstreamer_plugin.* GStreamer 'plugin' library file(s).
  • A license file.
  • This documentation in HTML form.

Please get in touch with our team to request the Plumerai People Detection demo package if you haven't done so yet. Note that it is merely designed as a demo for evaluation purposes, and not for production usage. For production, we offer our people detection library, as documented on docs.plumerai.com, which can be integrated into your application.

This demo uses the open-source GStreamer software to capture images from a camera and displays results on your screen. The GStreamer components are not optimized for speed in any way: the demo application might run significantly slower compared to actually using the Plumerai People Detection library in a production environment.

Apart from the GStreamer-based demo, we provide source code of a simple OpenCV webcam-to-display pipeline. This can be adjusted as needed for proto-typing and can also serve as a demo. Documentation and source code is available in a GitHub repository.

The GStreamer-based demo is pre-packaged for the following platforms:

  • Linux x86-64 with V4L. This will work on most Linux systems with most built-in webcams or USB cameras. See here how to set-up your system, here how to run it, and here for troubleshooting. Use this binary only if uname -a (in a terminal) includes x86_64.
  • Linux aarch64 with V4L. This will work on many embedded platforms with 64-bit ARM processors. See here how to set-up your system, here how to run it, and here for troubleshooting. Use this binary only if uname -a (in a terminal) includes aarch64.
  • Linux aarch64 with libcamera. This version is meant only for 64-bit ARM systems with Raspberry Pi's PiCam camera. For a RaspberryPi with another camera, use the version above with V4L. See here how to set-up your system, here how to run it, and here for troubleshooting. Use this binary only if uname -a (in a terminal) includes aarch64.
  • macOS x86-64 or arm64. This will work on most Apple computers with a built-in webcam or USB cameras. See here how to set-up your system, here how to run it, and here for troubleshooting.

The demo application can handle two forms of input:

  1. Live camera data, e.g. from a webcam. This is the default and recommended way to evaluate the software.
  2. Pre-recorded video files from disk. This is more experimental and might not work with all input formats.

The demo application can handle two forms of output:

  1. Output to screen. This is the default and recommended way to evaluate the software.
  2. Output to a video file on disk. This is more experimental and only recommended when your set-up does not support graphical output.