camera-management

camera-management

This is a low level way of interacting with multiple cameras connected to different devices. If you want to contribute please have a look at the DEV_README.

Installation

Use the following command to install the package:

pip install camera-management

If you want to use AVT-Cameras or any camera that uses the Vimba / VimbaX backend, please read the section about AVT Cameras below.

Usage

This package aims to provide an easy solution for communicating with multiple Webcams and other Cameras. A basic example is shown below, more examples can be found on GitHub:

Managed Usage

If you want to use the Manager to start all the camera Applications you can use the ManagerApp. This will start the desired camera backends, which can be accessed using the API-endpoints.

WARNING: If you want to have as little latency as possible and a high framerate consider using the standalone mode described below.

1. Set up and start the ManagerApp on the backend device, meaning the device with all the connected cams you want to use.

from camera_management.camera_management_app.manager_interface import ManagerApp
from pathlib import Path

manager_backend = ManagerApp(path_to_configs=Path(__file__).parent / "data", autostart=True)
manager_backend.start()

For a detailed explanation on how the configs work and how to build your own config see the section "configs" below. The config name can be anything you want.

1. Set up and start the manager_interface on the frontend device, meaning the device you want to use the camera stream at. You need to supply the IP address of the backend device. The port should always be 8090. If the front- and backend are the same device, the host IP will be 127.0.0.1.

from camera_management.frontends.basic_frontend import manager_interface

manager_frontend = manager_interface(host="127.0.0.1", port=8090, autostart=True)

1. Use the manager_frontend to get an actual camera interface. The manager only manages the cameras, but it does not provide any streams. It only provides the interfaces to the actual camera streams

# Get all ports of the actual configured cameras
print(camera_infos = manager_frontend.get_configured_cams())

The command above might print something like:

CAM TYPE               CAM SERIAL            CONFIG TYPE    CONFIG INDIVIDUAL      PORT  PORT ACTIVE    CALIBRATION AVAILABLE
---------------------  --------------------  -------------  -------------------  ------  -------------  -----------------------
4K USB CAMERA          12567                 True           False                  8091  True           False
webcam                 V011R007C001B004      True           False                  8092  True           False

Use this information to choose the camera you want to access.

# Get a camera interface by its port
camera_interface = manager_frontend.get_interface_by_port(8091)

1. Display the camera stream

while True:
    data = camera_interface.fetch_image_data()
    cv2.imshow("Stream", data.image)
    if cv2.waitKey(20) == ord("q"):
        break

The above while loop continuously polls the latest image from the camera backend and displays it using OpenCV. The function fetch_data() is a convenience function, the same result can be achieved using the code below, using requests.

import requests
import pickle
import cv2

host = '127.0.0.1'
port = 8090
while True:
    try:
        response = requests.get(f"http://{host}:{port}/data/imageProcessor")
    except requests.exceptions.ConnectionError:
        continue
    if response.status_code == 200:
        cv2.imshow("Moin", pickle.loads(response.content).image)
        if cv2.waitKey(20) == ord("q"):
            break

1. Change camera settings. You can use the convenience function set_general_setting to set camera settings. Since this library works for a range of vastly different cameras, not every setting can be applied to each camera. Check your specific camera to see, if a specific settings is applicable. Below is an example of setting the CAP_PROP_FRAME_WIDTH property, meaning the width of the output image to 400 px. Afterward, the settings is read using get_general_setting. Finally, the current active camera description is shown using the description property of the camera interface.

Important: When using the convenience function set_general_setting, the setting is automatically read back, so you can see if it was successful.

from camera_management import SETTINGS
print(camera_interface.set_general_setting(SETTINGS.CAP_PROP_FRAME_WIDTH, 400))
print(camera_interface.get_general_setting(SETTINGS.CAP_PROP_FRAME_WIDTH))
print(camera_interface.description)

Standalone Mode

If you want to achieve a higher FPS and lower latency you should use the standalone mode.

Description Files

A valid camera description file can be created using the following code. It will create description files for every currently connected cam and save them to the supplied directory.

from pathlib import Path
from camera_management.tools.create_description import create_basic_description

create_basic_description(Path('<PATH_WHERE_CONFIG_SHOULD_BE_STORED>'))

This will create valid, yet empty description files in the given directory. A valid description file might look like this:

{
  "information": {
    "standard_resolution": {
      "x": 1920,
      "y": 1080,
      "channels": 3
    },
    "available_resolutions": [
      {
        "x": 1920,
        "y": 1080,
        "channels": 3
      }
    ],
    "device": {
      "path": 0,
      "vendor_id": "0x4f2",
      "product_id": "0xb79e",
      "vendor": "SunplusIT Inc",
      "product": "USB2.0 FHD UVC WebCam",
      "serial": null,
      "backend": 1200,
      "transp_type": "usb ",
      "unique_id": "0x14000004f2b79e",
      "device_type": "AVCaptureDeviceTypeExternalUnknown",
      "is_used": false
    },
    "resolution_roi_coupled": false
  },
  "calibration": {
    "mode": "type",
    "available": false,
    "values": null,
    "model": null
  },
  "config": {},
  "debug": {
    "last_updated": "22-01-2024 16:35:47",
    "created_on_platform": "darwin"
  }
}

• information: Contains general information about a camera
  • standard_resolution: Contains the desired resolution of the output stream
  • available_resolutions: Contains all available resolutions of the camera
  • device: Contains all the relevant information to identify a camera type or an individual camera.
  • resolution_roi_coupled: Is the resolution coupled to the ROI (region of interest)? Can be different for different cameras
• calibration: Calibration information and values. Can be a type or individual calibration.
  • mode: A type calibration is a set of general calibration parameters for a certain camera type, where as an individual calibration is linked to a specific (individual) camera
  • available: Are there calibration values?
  • values: The actual calibration values. Should be empty when 'available' is set to False and should contain values, if 'available' is set to True. The actual values can be different, depending on the chosen calibration model
  • model: The chosen calibration model (can be 'OpenCV_standard' or 'brown' at the moment)
• config: The initial camera settings. Leave empty if you don't want to set them. This field is dict containing initial camera settings given as a key: value pair (e.g. '3: 400' -> This sets the image width to 400)
• debug: Information to debug a description file (Please don't change those manually)
  • last_updated: The last time this file was updated
  • created_on_platform: The platform this file was created on

Known Issues

• When using Cameras with a large resolution the framerate might be very low
• When using multiple USB 2.x cameras connected to the manager device using an USB Hub some might not work, since the data rate of USB 2.x is limited.
• When using the create_basic_description function it will create a parseable description file which is filled with standard paramters that should be adjusted after creation

AVT Cameras

AVT (Allied Vision Technologies) Cameras or all other cameras that use the Vimba or VimbaX backend are supported in this project, but need to be set up manually, since the Vimba / Vimba X python wrapper currently does not sit on pypi. Vimba X is currently the most recent version of Vimba. Although it shares many features with Vimba it is a standalone release. The python wrapper for Vimba X is called VmbPy (not to be confused with Vimba Python, which is the Python wrapper for Vimba - without X). To install VmbPy you have to follow the instructions here.

Warning: VmbaPy is not very stable at the moment and does require some tweaks to run on Linux with Ethernet Cams.