This project takes images and then performs monodepth estimation using the OpenVINO toolkit and the Intel NCS2
The bin
file was too big to add so use this link to download the file and add it to your model
directory.
To get started, you will need the following:
- Raspberry Pi 4 (1, 2, 4, or 8GB)
- MicroSD Card (At least > 16GB)
- USB-C Power
- SSH Enabled on the Pi for remote access
- Intel NCS2 for Inferencing
- OpenVINO toolkit (follow this gist for instructions)
- Raspberry Pi Camera (Any will suffice)
- Camera Mount (recommended if using outdoors)
This project was tested using Raspberry Pi OS Bullseye 64-bit. Once I get my hands on the Raspberry Pi 5 with Bookworm I will run tests. To run this application just type python3 app.py
and then go to ip-address-of-pi:5000
and the app should appear as shown below.
Take a picture and then run monodepth estimation on the image. A sample image is shown below.