FrameGrab by Groundlight

A user-friendly library for grabbing images from cameras or streams

FrameGrab is an open-source Python library designed to make it easy to grab frames (images) from cameras or streams. The library supports generic USB cameras (such as webcams), RTSP streams, Basler USB cameras, Basler GigE cameras, Intel RealSense depth cameras, and video file streams (mp4, mov, mjpeg, avi, etc.).

FrameGrab also provides basic motion detection functionality. FrameGrab requires Python 3.7 or higher.

Table of Contents

• FrameGrab by Groundlight
  • A user-friendly library for grabbing images from cameras or streams
  • Table of Contents
  • Installation
  • Optional Dependencies
  • Usage
    • Command line interface (CLI)
    • Frame Grabbing
    • Configurations
    • Autodiscovery
      • RTSP Discovery
    • Motion Detection
  • Examples
    • Generic USB
    • YouTube Live
    • File Stream
  • Contributing
  • License

Installation

To install the FrameGrab library, simply run:

pip install framegrab

Optional Dependencies

Certain camera types have additional dependencies that must be installed separately. If you don't intend to use these camera types, you don't need to install these extra packages.

• To use a Basler USB or GigE camera, you must separately install the pypylon package.
• To use Intel RealSense cameras, you must install pyrealsense2.
• To use a Raspberry Pi "CSI2" camera (connected with a ribbon cable), you must install the picamera2 library. See install instructions at the picamera2 github repository.
• To use a YouTube Live stream, you must install streamlink.

We provide optional extras to install these dependencies. For example, to install the Basler camera dependencies, run:

pip install framegrab[basler]

To install YouTube Live stream dependencies, run:

pip install framegrab[youtube]

To install all optional dependencies, run:

pip install framegrab[all]

Usage

Command line interface (CLI)

There is a simple CLI for framegrab to discover and preview configurations.

framegrab

lists the sub-commands, including autodiscover and preview.

Frame Grabbing

Frame Grabbers are defined by a configuration dict which is usually stored as YAML. The configuration combines the camera type, the camera ID, and the camera options. The configuration is passed to the FrameGrabber.create_grabber method to create a grabber object. The grabber object can then be used to grab frames from the camera.

config can contain many details and settings about your camera, but only input_type is required. Available input_type options are: generic_usb, rtsp, realsense, basler, rpi_csi2, hls, youtube_live, and file_stream.

Here's an example of a single USB camera configured with several options:

config = """
name: Raspberry Pi Ribbon Cable Camera
input_type: rpi_csi2
options:
    resolution:
        height: 720
        width: 1280
    zoom:
        digital: 1.5
"""

grabber = FrameGrabber.create_grabber_yaml(config)

To get a frame, simply run:

frame = grabber.grab()

You can also change the options after the grabber is created.

new_options = {
    'resolution': {
        'height': 480,
        'width': 640,
    },
    'crop': {
        'relative': {
            'top': .1,
            'bottom': .9,
            'left': .1,
            'right': .9,
        }
    }
}

grabber.apply_options(new_options)

When you are done with the camera, release the resource by running:

grabber.release()

Alternatively, you can use a context manager which will automatically release the camera resources:

with FrameGrabber.create_grabber_yaml(config) as grabber:
    frame = grabber.grab()

You might have several cameras that you want to use in the same application. In this case, you can load the configurations from a yaml file and use FrameGrabber.create_grabbers. Note that currently only a single Raspberry Pi CSI2 camera is supported, but these cameras can be used in conjunction with other types of cameras.

If you have multiple cameras of the same type plugged in, it's recommended that you include serial numbers in the configurations; this ensures that each configuration is paired with the correct camera. If you don't provide serial numbers in your configurations, configurations will be paired with cameras in a sequential manner.

Below is a sample yaml file containing configurations for three different cameras.

image_sources:
  - name: On Robot Arm
    input_type: basler
    id:
      serial_number: A24P1V4T
    options:
      crop:
        relative:
          top: 0.3
          right: 0.8
  - name: Chip Bin
    input_type: rtsp
    id:
      rtsp_url: rtsp://admin:[email protected]/cam/realmonitor?channel=1&subtype=0
    options:
      crop:
        pixels:
          top: 350
          bottom: 1100
          left: 1100
          right: 2000
  - name: Over CNC Machine
    input_type: generic_usb
    id:
      serial_number: B77D3A8F

You can load the configurations from the yaml file and use the cameras in the following manner.

from framegrab import FrameGrabber

config_path = 'camera_config.yaml'
grabbers = FrameGrabber.from_yaml(config_path)

for grabber in grabbers.values():
    print(grabber.config)
    frame = grabber.grab()
    display_image(frame) # substitute this line for your preferred method of displaying images, such as cv2.imshow
    grabber.release()

Configurations

The table below shows all available configurations and the cameras to which they apply.

Configuration Name	Example	Generic USB	RTSP	Basler	Realsense	Raspberry Pi CSI2	HLS	YouTube Live	File Stream
name	On Robot Arm	optional	optional	optional	optional	optional	optional	optional	optional
input_type	generic_usb	required	required	required	required	required	required	required	required
id.serial_number	23458234	optional	-	optional	optional	-	-	-	-
id.rtsp_url	rtsp://…	-	required	-	-	-	-	-	-
id.hls_url	https://.../*.m3u8	-	-	-	-	-	required	-	-
id.youtube_url	https://www.youtube.com/watch?v=...	-	-	-	-	-	-	required	-
id.filename	http://.../*.mp4	-	-	-	-	-	-	-	required
options.resolution.height	480	optional	-	-	optional	-	-	-	-
options.resolution.width	640	optional	-	-	optional	-	-	-	-
options.zoom.digital	1.3	optional	optional	optional	optional	optional	optional	optional	optional
options.crop.pixels.top	100	optional	optional	optional	optional	optional	optional	optional	optional
options.crop.pixels.bottom	400	optional	optional	optional	optional	optional	optional	optional	optional
options.crop.pixels.left	100	optional	optional	optional	optional	optional	optional	optional	optional
options.crop.pixels.right	400	optional	optional	optional	optional	optional	optional	optional	optional
options.crop.relative.top	0.1	optional	optional	optional	optional	optional	optional	optional	optional
options.crop.relative.bottom	0.9	optional	optional	optional	optional	optional	optional	optional	optional
options.crop.relative.left	0.1	optional	optional	optional	optional	optional	optional	optional	optional
options.crop.relative.right	0.9	optional	optional	optional	optional	optional	optional	optional	optional
options.depth.side_by_side	1	-	-	-	optional	-	-	-	-
options.num_90_deg_rotations	2	optional	optional	optional	optional	optional	optional	optional	optional
options.keep_connection_open	True	-	optional	-	-	-	optional	optional	-
options.max_fps	30	-	optional	-	-	-	-	-	optional

In addition to the configurations in the table above, you can set any Basler camera property by including options.basler.<BASLER PROPERTY NAME>. For example, it's common to set options.basler.PixelFormat to RGB8.

Autodiscovery

Autodiscovery automatically connects to cameras that are plugged into your machine or discoverable on the network, including generic_usb, realsense, basler, and ONVIF supported rtsp cameras. Note that rpi_csi2 cameras are not yet supported by autodiscover. Default configurations will be loaded for each camera. Note that discovery of RTSP cameras will be disabled by default but can be enabled by setting rtsp_discover_mode. Refer to RTSP Discovery section for different options.

Autodiscovery is great for simple applications where you don't need to set any special options on your cameras. It's also a convenient method for finding the serial numbers of your cameras (if the serial number isn't printed on the camera).

grabbers = FrameGrabber.autodiscover()

# Print some information about the discovered cameras
for grabber in grabbers.values():
    print(grabber.config)

    grabber.release()

RTSP Discovery

RTSP cameras with support for ONVIF can be discovered on your local network in the following way:

from framegrab import RTSPDiscovery, ONVIFDeviceInfo

devices = RTSPDiscovery.discover_onvif_devices()

The discover_onvif_devices() will provide a list of devices that it finds in the ONVIFDeviceInfo format. An optional mode auto_discover_mode can be used to try different default credentials to fetch RTSP URLs:

• off: No discovery.
• ip_only: Only discover the IP address of the camera.
• light: Only try first two usernames and passwords ("admin:admin" and no username/password).
• complete_fast: Try the entire DEFAULT_CREDENTIALS without delays in between.
• complete_slow: Try the entire DEFAULT_CREDENTIALS with a delay of 1 seconds in between.

After getting the list and enter the username and password of the camera. Use generate_rtsp_urls() to generate RTSP URLs for each devices.

for device in devices:
    RTSPDiscovery.generate_rtsp_urls(device=device)

This will generate all the available RTSP URLs and can be used when creating FrameGrabber.create_grabbers to grab frames.

config = f"""
name: Front Door Camera
input_type: rtsp
id:
  rtsp_url: {device.rtsp_urls[0]}
"""

grabber = FrameGrabber.create_grabber_yaml(config)

Motion Detection

To use the built-in motion detection functionality, first create a MotionDetector object, specifying the percentage threshold for motion detection:

from framegrab import MotionDetector

motion_threshold = 1.0
m = MotionDetector(pct_threshold=motion_threshold)

The motion threshold is defined as the detection threshold for motion detection, in terms of the percentage of changed pixels. The default value is 1.0 (which means 1%).

Then, use the motion_detected() method with a captured frame to check if motion has been detected:

if m.motion_detected(frame):
    print("Motion detected!")

Examples

Generic USB

Here's an example of using the FrameGrab library to continuously capture frames and detect motion from a video stream:

from framegrab import FrameGrabber, MotionDetector

motion_threshold = 1.0
m = MotionDetector(pct_threshold=motion_threshold)

config = {
    'input_type': 'generic_usb',
}

with FrameGrabber.create_grabber(config) as grabber:
    while True:
        frame = grabber.grab()
        if frame is None:
            print("No frame captured!")
            continue

        if m.motion_detected(frame):
            print("Motion detected!")

YouTube Live

Here's an example of using FrameGrab to capture frames from a YouTube Live stream:

from framegrab import FrameGrabber
import cv2

config = {
    'input_type': 'youtube_live',
    'id': {
        'youtube_url': 'https://www.youtube.com/watch?v=your_video_id'
    }
}

with FrameGrabber.create_grabber(config) as grabber:
    frame = grabber.grab()
    if frame is None:
        raise Exception("No frame captured")

    # Process the frame as needed
    # For example, display it using cv2.imshow()
    # For example, save it to a file
    cv2.imwrite('youtube_frame.jpg', frame)

File Stream

Here's an example of using FrameGrab to capture frames from a video file:

from framegrab import FrameGrabber
import cv2

config = {
    'input_type': 'file_stream',
    'id': {
        'filename': 'path/to/your/video.mjpeg'  # or .mp4, .avi, .mov, etc.
    },
    'options': {
        'max_fps': 2,  # if a lower FPS than the original video's FPS is specified, Framegrab will skip extra frames as needed.
    }
}

with FrameGrabber.create_grabber(config) as grabber:
  frame = grabber.grab()
  if frame is None:
      raise Exception("No frame captured")

  # Process the frame as needed
  # For example, display it using cv2.imshow()
  # For example, save it to a file
  cv2.imwrite('file_stream_frame.jpg', frame)

Contributing

We welcome contributions to FrameGrab! If you would like to contribute, please follow these steps:

1. Fork the repository
2. Create a new branch for your changes
3. Commit your changes to the branch
4. Open a pull request

License

FrameGrab is released under the MIT License. For more information, please refer to the LICENSE.txt file.