RaspiBallPos

Installation

1. Enable Raspberry Camera module by running sudo raspi-config in case you haven't already done it.
2. Install the required Python modules by running sudo pip3 install picamera click flask matplotlib profilehooks screeninfo imutils RPi.GPIO flask_bootstrap flask_colorpicker scipy
3. Install the eigen library by sudo apt install libeigen3-dev
4. Clone this repository by git clone https://github.com/aa4cc/raspi-ballpos.git
5. Compile the necessary modules by running ./compile_libraries
6. Go back by cd ../..
7. Run cp raspi-ballpos/config.json_sample config.json
8. Edit your configuration using your favourite editor (do not forget to specify screen resolution, otherwise the detected object position will not be displayed correctly).
9. Run cd raspi-ballpos
10. Run python3 vision.py -vp and see the result.

Automatic startup

1. In order to start the process automatically upon booting, copy vision.service into /etc/systemd/system/ (run sudo cp vision.service /etc/systemd/system/vision.service).
2. Open the file (sudo nano /etc/systemd/system/vision.service) and change ExecStart and WorkingDirectory paths to vision.py and raspi-ballpos respectively and optionally set flags for starting.
3. Use sudo systemctl ACTION vision with actions such as start, stop, enable, disable to control the service.

Usage

Usage: vision.py [OPTIONS]

Options:

Shortcut	Option	Type	Description
-c	--config-file	FILE PATH	Path to config file (default: ../config.json)
-e	--exposition-time	INTEGER	Exposition time (shutter speed) in milliseconds.
-v	--verbose	flag	Display time needed for processing of each frame and the measured positions.
-p	--preview	flag	Show preview on HDMI or display.
-l	--load-old-color-settings	flag	Loads old color settings if using MultiColorDetector.
	--video-record	flag	Record video.
	--img-path	TEXT	Path to store images and videos ideally ramdisk.
	--interactive	flag	Start interactive Python console, to get realtime access to PiCamera object for testing purposes.
	--multicore	flag	Start detectors in different processes to speed-up detection.
	--web-interface/--no-web-interface	flag	Enable/Disable web interface on port 5001 (default: enable).
	--help	flag	Show this message and exit.

Web interface

If web interface is enabled, it can be accessed by IP address or hostname on port 5001. All code for webinterface is based in interface.py file.

Example: HTTP://yourhostname.com:5001

URL endpoints:

• / Summary page - Print all used settings and live images for all decoders (if not using MultiColorDetector).
• /detector/NAME - Print settings and live images for detector called NAME.
• /centers - Returns JSON with positions of all objects.
• /config - Returns live config in JSON which can be used to replicates settings on another instance in config.json (for changes to MultiColorDetector, see /ball_colors).
• /config - POST request loads new config form request into live config and restarts detection subsystem.
• /config/loadfile - POST request loads new config from local filesystem by given filename into live config and restarts detection subsystem.
• /detector/NAME/threshold/VALUE - Sets threshold of detector NAME to VALUE (for changes to MultiColorDetector, see /ball_colors).
• /image - Returns the latest PNG color image.
• /image/DETECTOR/TYPE - Returns the latest PNG image from detector DETECTOR and TYPE chosen from image, image_dwn,downsample,downsample_thrs,roi,roi_thrs or im_thrs for MultiColorDetector.
• /imagesc/DETECTOR/TYPE - Same as /image/DETECTOR/TYPE but rendered by matplotlib imshow.
• /wb - Page for white balancing camera.
• /lamp/on, /lamp/off - Turn lamp on and off respectively (if available).
• /restart - Restart the image detection subsystem.
• /ball_colors - UI for live color settings, only works with MultiColorDetector/RANSAC. Example of how to properly configure such a detector can be found in config_multi.json. Use GET option i=# (i.e. URL /ball_colors?i=100) to get a custom number of frames from which to calculate mean and standard deviation (default 5).
• /triangle - Shows the perceived triangle along with its centroid and all ball centers. Currently, balls used for triangle are hard-coded as 1, 4, 5.
• /ransac - Ransac settings (see the manual for more info)

Additional notes

• If you're controlling RPi remotely over SSH and wish to view the preview on a connected screen, you need to set export DISPLAY=:0.0. https://stackoverflow.com/questions/13046624/how-to-permanently-export-a-variable-in-linux