Home

Awesome

DepthAI API Demo Program

Discord Forum Docs License: MIT

This repo contains demo application, which can load different networks, create pipelines, record video, etc.

Click on the GIF below to see a full example run

depthai demo

Documentation is available at https://docs.luxonis.com/en/latest/pages/tutorials/first_steps/.

Installation

First you need to clone this repository with

git clone --recursive https://github.com/luxonis/depthai.git

In case you have repository already cloned, you can update your submodules with:

git pull --recurse-submodules 

There are two installation steps that need to be performed to make sure the demo works:

Docker

One may start any DepthAI programs also through Docker: (Allowing X11 access from container might be required: xhost local:root)

DepthAI Demo

docker run --privileged -v /dev/bus/usb:/dev/bus/usb --device-cgroup-rule='c 189:* rmw' -e DISPLAY=$DISPLAY -v /tmp/.X11-unix:/tmp/.X11-unix --network host --rm -i -t luxonis/depthai:latest python3 /depthai/depthai_demo.py

Calibration

docker run --privileged -v /dev/bus/usb:/dev/bus/usb --device-cgroup-rule='c 189:* rmw' -e DISPLAY=$DISPLAY -v /tmp/.X11-unix:/tmp/.X11-unix --network host --rm -i -t luxonis/depthai:latest python3 /depthai/calibrate.py [parameters]

Usage

This repository and the demo script itself can be used in various independent cases:

QT GUI usage

See instuctions here

qt demo

Command line usage

Examples

python3 depthai_demo.py -gt cv - RGB & CNN inference example

python3 depthai_demo.py -gt cv -vid <path_to_video_or_yt_link> - CNN inference on video example

python3 depthai_demo.py -gt cv -cnn person-detection-retail-0013 - Runs person-detection-retail-0013 model from resources/nn directory

python3 depthai_demo.py -gt cv -cnn tiny-yolo-v3 -sh 8 - Run tiny-yolo-v3 model from resources/nn directory and compile for 8 shaves

Demo

cv demo

For the full reference, run $ depthai_demo.py --help.

DepthAI Apps

We currently have 2 apps, uvc and record.

UVC app

This app will upload an UVC pipeline to the connected OAK camera, so you will be able to use an OAK as a webcam.

Record App

Record app lets you record encoded and synced video streams (MJPEG/H265) across all available devices into .mp4, Foxglove's .MCAP, or ROS' .bag . Since mono streams are synced, you will be able to reconstruct the whole stereo depth perception.

Run using $ depthai_demo.py -app record [-p SAVE_PATH] [-q QUALITY] [--save STREAMS] [-fc] [-tl]. More information about the arguments and replaying can be found here.

Supported models

We have added support for a number of different AI models that work (decoding and visualization) out-of-the-box with the demo. You can specify which model to run with -cnn argument, as shown above. Currently supported models:

- deeplabv3p_person
- face-detection-adas-0001
- face-detection-retail-0004
- human-pose-estimation-0001
- mobilenet-ssd
- openpose2
- pedestrian-detection-adas-0002
- person-detection-retail-0013
- person-vehicle-bike-detection-crossroad-1016
- road-segmentation-adas-0001
- tiny-yolo-v3
- vehicle-detection-adas-0002
- vehicle-license-plate-detection-barrier-0106
- yolo-v3

If you would like to use a custom AI model, see documentation here.

Usage statistics

By default, the demo script will collect anonymous usage statistics during runtime. These include:

We gather this data to better understand what environemnts are our users using, as well as assist better in support questions.

All of the data we collect is anonymous and you can disable it at any time. To do so, click on the "Misc" tab and disable sending the statistics or create a .consent file in repository root with the following content

{"statistics": false}

Reporting issues

We are actively developing the DepthAI framework, and it's crucial for us to know what kind of problems you are facing. If you run into a problem, please follow the steps below and email support@luxonis.com:

  1. Run log_system_information.sh and share the output from (log_system_information.txt).
  2. Take a photo of a device you are using (or provide us a device model)
  3. Describe the expected results;
  4. Describe the actual running results (what you see after started your script with DepthAI)
  5. How you are using the DepthAI python API (code snippet, for example)
  6. Console output