VidGear is a powerful python Video Processing library built with multi-threaded Gears(a.k.a APIs) each with a unique set of trailblazing features. These APIs provides a easy-to-use, highly extensible, and multi-threaded wrapper around many underlying state-of-the-art python libraries such as OpenCV ➶, FFmpeg ➶, picamera ➶, pafy ➶, pyzmq ➶ and python-mss ➶
The following functional block diagram clearly depicts the functioning of VidGear library:
Table of Contents
VidGear is an ultrafast➶, compact, flexible and easy-to-adapt complete Video Processing Python Library.
Built with simplicity in mind, VidGear lets programmers and software developers to easily integrate and perform complex Video Processing tasks in their existing or new applications, without going through various underlying python library's documentation and using just a few lines of code. Beneficial for both, if you're new to Programming with Python language or a pro at it.
For more advanced information see the Wiki Documentation ➶.
New Release SneekPeak : VidGear 0.1.5
- Released new ScreenGear API, supports Live ScreenCasting.
- Released new NetGear API, aids real-time frame transfer through messaging(ZmQ) over the network.
- Released new Stabilizer Class, for minimum latency Video Stabilization with OpenCV.
- Updated VideoGear API to be used as an internal wrapper around Stabilizer Class.
- Implemented exclusive Threaded Queue Mode for blazingly fast, synchronized and error-free multi-threading APIs.
- Added Option to use VidGear API's standalone.
- Several Performance enhancements and Bugs exterminated.
- Revamped Docs and many more...
To use VidGear in your python application, you must check the following dependencies before you install VidGear :
OpenCV:VidGear must require OpenCV(3.0+) python enabled binaries to be installed on your machine for its core functions. For its installation, you can follow these online tutorials for linux and raspberry pi, otherwise, install it via pip:
pip install opencv-python
FFmpeg:VidGear must require FFmpeg for its powerful video compression and encoding capabilities.
🌟Follow this FFmpeg wiki page for its installation. 🌟
picamera:Required for using Raspberry Pi Camera Modules(such as OmniVision OV5647 Camera Module) on your Raspberry Pi machine. You can easily install it via pip:
pip install picamera
Also, make sure to enable Raspberry Pi hardware-specific settings prior to using this library.
mss:Required for Screen Casting. Install it via pip:
pip install mss
pyzmq:Required for transferring video frames through ZeroMQ messaging system over the network. Install it via pip:
pip install pyzmq
pip install pafy pip install -U youtube-dl
Option 1: PyPI Install
Best option for quickly getting VidGear installed.
pip install vidgear
Option 2: Release Archive Download
Best option if you want a compressed archive.
VidGear releases are available for download as packages in the latest release
Option 3: Clone the Repository
Best option for automatically installing required dependencies(except FFmpeg), or for latest patches(maybe experimental), or contributing to development.
You can clone this repository's
testing branch for development and thereby can install as follows:
git clone https://github.com/abhiTronix/vidgear.git cd vidgear git checkout testing pip install .
VidGear is built with multi-threaded Gears each with some unique function/mechanism. Each Gear is designed exclusively to handle/control different device-specific video streams, network streams, and media encoders. These APIs provides an easy-to-use, highly extensible, and a multi-threaded wrapper around many underlying various python libraries to exploit their features and functions directly while providing robust error-handling.
These Gears can be classified as follows:
A. VideoCapture Gears:
- CamGear: Targets various IP-USB-Cameras/Network-Streams/YouTube-Video-URL.
- PiGear: Targets various Raspberry Pi Camera Modules.
- ScreenGear: Enables ultra-fast Screen Casting.
- VideoGear: A common API with Video Stabilizer wrapper.
B. VideoWriter Gear:
- WriteGear: Handles easy Lossless Video Encoding and Compression.
C. Network Gear:
- NetGear: Targets synchronous video frames transferring between interconnecting systems over the network.
CamGear supports a diverse range of video streams which can handle/control video stream almost any IP/USB Cameras, multimedia video file format (upto 4k tested), network stream URL such as
http(s), rtp, rstp, mms, etc. In addition to this, it also supports live Gstreamer's RAW pipelines and YouTube video/livestreams URLs. CamGear provides a flexible, high-level multi-threaded wrapper around
OpenCV's VideoCapture class with access almost all of its available parameters and also employs
pafy's APIs for live YouTube streaming. Furthermore, CamGear relies exclusively on Threaded Queue mode for ultra-fast, error-free and synchronized frame handling.
Following simplified functional block diagram depicts CamGear API's generalized working:
CamGear API Guide:
VideoGear API provides a special internal wrapper around VidGear's exclusive Video Stabilizer class. Furthermore, VideoGear API can provide internal access to both CamGear and PiGear APIs separated by a special flag. Thereby, this API holds the exclusive power for any incoming VideoStream from any source, whether it is live or not, to stabilize it directly with minimum latency and memory requirements.
Below is a snapshot of a VideoGear Stabilizer in action:
Original Video Courtesy@SIGGRAPH2013
Code to generate above VideoGear API Stabilized Video(See more detailed usage examples here):
# import libraries from vidgear.gears import VideoGear import numpy as np import cv2 stream_stab = VideoGear(source='test.mp4', stabilize = True).start() # To open any valid video stream with `stabilize` flag set to True. stream_org = VideoGear(source='test.mp4').start() # open same stream without stabilization for comparison # infinite loop while True: frame_stab = stream_stab.read() # read stabilized frames # check if frame is None if frame_stab is None: #if True break the infinite loop break #read original frame frame_org = stream_org.read() #concatenate both frames output_frame = np.concatenate((frame_org, frame_stab), axis=1) #put text cv2.putText(output_frame, "Before", (10, output_frame.shape - 10),cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0,255,0), 2) cv2.putText(output_frame, "After", (output_frame.shape//2+10, frame.shape - 10),cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0,255,0), 2) cv2.imshow("Stabilized Frame", output_frame) # Show output window key = cv2.waitKey(1) & 0xFF # check for 'q' key-press if key == ord("q"): #if 'q' key-pressed break out break cv2.destroyAllWindows() # close output window stream_org.stop() stream_stab.stop() # safely close video streams.
VideoGear API Guide:
PiGear is similar to CamGear but made to support various Raspberry Pi Camera Modules (such as OmniVision OV5647 Camera Module and Sony IMX219 Camera Module). To interface with these modules correctly, PiGear provides a flexible multi-threaded wrapper around complete picamera python library, and provides us the ability to exploit its various features like
brightness, saturation, sensor_mode, etc. effortlessly.
Following simplified functional block diagram depicts PiGear API:
PiGear API Guide:
With ScreenGear, we can easily define an area on the computer screen or an open window to record the live screen frames in real-time at the expense of insignificant latency. To achieve this, ScreenGear provides a high-level multi-threaded wrapper around
mss python library API and also supports the flexible direct parameter manipulation.
Below is a snapshot of a ScreenGear API in action:
Code to generate the above result:
# import libraries from vidgear.gears import ScreenGear import cv2 stream = ScreenGear().start() # infinite loop while True: frame = stream.read() # read frames # check if frame is None if frame is None: #if True break the infinite loop break cv2.imshow("Output Frame", frame) # Show output window key = cv2.waitKey(1) & 0xFF # check for 'q' key-press if key == ord("q"): #if 'q' key-pressed break out break cv2.destroyAllWindows() # close output window stream.stop() # safely close video stream.
ScreenGear API Guide:
WriteGear is undoubtedly the most powerful Video Processing Gear of them all. It solely handles various powerful FFmpeg tools that allow us to do almost anything you can imagine with multimedia files. With WriteGear API, you can process real-time video frames into a lossless format and specification suitable for our playback in just a few lines of codes. These specifications include setting bitrate, codec, framerate, resolution, subtitles, compression, etc. Furthermore, we can multiplex extracted audio at the output with compression and all that in real-time(see this example). In addition to this, WriteGear also provides flexible access to OpenCV's VideoWriter API which provides some basic tools for video frames encoding but without compression.
WriteGear primarily operates in the following two modes:
Compression Mode: In this mode, WriteGear utilizes
FFmpeg'sinbuilt encoders to encode lossless multimedia files. It provides us the ability to exploit almost any available parameters available within FFmpeg, with so much ease and flexibility and while doing that it robustly handles all errors/warnings quietly. You can find more about this mode here.
Non-Compression Mode: In this mode, WriteGear utilizes basic OpenCV's inbuilt VideoWriter API. Similar to compression mode, WriteGear also supports all parameters manipulation available within OpenCV's VideoWriter API. But this mode lacks the ability to manipulate encoding parameters and other important features like video compression, audio encoding, etc. You can learn about this mode here.
Following functional block diagram depicts WriteGear API's generalized working:
WriteGear API Guide:
NetGear is exclusively designed to transfer video frames synchronously between interconnecting systems over the network in real-time. This is achieved by implementing a high-level wrapper around PyZmQ python library that contains python bindings for ZeroMQ - a high-performance asynchronous distributed messaging library that aim to be used in distributed or concurrent applications. It provides a message queue, but unlike message-oriented middleware, a ZeroMQ system can run without a dedicated message broker. Furthermore, NetGear currently supports two ZeroMQ messaging patterns: i.e
zmq.REQ and zmq.REP.
NetGear API has two modes of operations:
- Send Mode:(a.k.a Server configuration) which employs
send()function to send frames to the client(s). You can use this function to send messages to client(s) too.
- Receive Mode:(a.k.a Client configuration) which employs
recv()function to receive frames sent by server. Client send back confirmation when frame is received successfully.
Following functional block diagram depicts NetGear API:
NetGear API Guide:
The full documentation for all VidGear classes and functions can be found in the link below:
Prerequisites: Testing VidGear require some additional dependencies & data which can be downloaded manually as follows:
Clone & Install Testing Branch
Download few additional python libraries:
pip install six pip install pytest
Download Test Dataset: To perform tests, additional test dataset is required, which can be downloaded by running bash script as follows:
chmod +x scripts/prepare_dataset.sh ./scripts/prepare_dataset.sh #for windows, use `sh scripts/pre_install.sh`
Run Tests: Then various VidGear tests can be run with
pytest(in VidGear's root folder) as below:
pytest -sv #-sv for verbose output.
Supported Python legacies
Python 2.7 legacies: VidGear v0.1.5 is officially the last Python 2.7 legacies supporting version. Kindly migrate your source code to Python 3 as soon as possible.
Python 3.x legacies: follows the numpy releases.
Copyright © abhiTronix 2019
This project is licensed under the MIT license.