Object Detection using OpenCV

The application demonstrates real-time object detection using a USB camera and is implemented in both C++ and Python.

Prerequisites

Synaptics Astra SL1680 evaluation kit with either OOBE or Non-OOBE image installed
USB camera connected to the board
HDMI display (for viewing results)
For Python implementation (Internet connectivity for initial setup)

Non-OOBE Implementation

The Non-OOBE image includes only the C++ implementation of the object detection application.

Using C++ Implementation

Install a pre-built image and connect to the board as described here
Connect the USB camera to the board
Set the Wayland display environment variable:
```
$ export WAYLAND_DISPLAY=wayland-1
```
Run the OpenCV object detection application:
```
$ opencv-gst-od
```

../_images/opencv_od_app_cpp_execution_non_oobe.png — OpenCV Object Detection Application Running (C++) in Non-OOBE Image

The application will display the live camera feed with detected objects highlighted on the connected display.

OOBE Implementation

The OOBE image includes both C++ and Python implementations of the object detection application.

Using C++ Implementation

Install a pre-built image and connect to the board as described here
Connect the USB camera to the board
Set the Wayland display environment variable:
```
$ export WAYLAND_DISPLAY=wayland-1
```
Run the OpenCV object detection application:
```
$ opencv-gst-od
```

../_images/opencv_od_app_cpp_execution.png — OpenCV Object Detection Application Running (C++) in OOBE Image

Using Python Implementation

The Python implementation provides an alternative way to run object detection. This implementation requires internet connectivity for initial setup.

Install a pre-built image and connect to the board as described here
Connect the USB camera to the board
Ensure the board has internet connectivity
Navigate to the demos directory:
```
$ cd /home/root/demos/scripts/
```

Create and activate a Python virtual environment with system packages:

$ python3 -m venv .venv --system-site-packages
$ source .venv/bin/activate

Install the Synaptics Python package:
```
$ pip3 install synap-python
```
Run the Python object detection application:
```
$ python3 sl1680_od_cam.py
```

../_images/opencv_od_app_python_execution.png — OpenCV Object Detection Application Running (Python) in OOBE Image

Note

The Python implementation requires internet connectivity during the initial setup to download the synap-python package and any required dependencies. After setup, the application can run offline.

Note

Make sure to activate the virtual environment (step 5) before running the Python application each time you want to use it.

Troubleshooting

Camera Not Detected

If the USB camera is not detected, ensure it is properly connected and recognized by the system:

Get list of video devices to verify the camera is detected::

$ v4l2-ctl --list-devices

No Display Output

If the video is not displaying, verify the Wayland display variable is set correctly:

$ echo $WAYLAND_DISPLAY

It should output wayland-1. If it’s empty, set it again using:

$ export WAYLAND_DISPLAY=wayland-1

Python Module Not Found

If you encounter module import errors when running the Python implementation, ensure the virtual environment is activated:

$ source /home/root/demos/scripts/.venv/bin/activate

Then verify that synap-python is installed:

$ pip3 list | grep synap-python