Mydia 1.0.4

Read videos as numpy arrays

Reading videos into NumPy arrays was never more simple. In addition, this library also provides an entire range of additional functionalities for reading the videos.

Read the Documentation

Getting started

How to simple read a video, given its path?

# Import
from mydia import Videos

# Initialize video path
video_path = r'./static/sample_video/bigbuckbunny.mp4'

# Create a reader object
reader = Videos()

# Call the 'read()' function to get the video tensor
video = reader.read(video_path)   # a tensor of shape (1, 132, 720, 1080, 3)

The tensor represents 1 video having 132 frames, with each frame having a width and height of 1080 and 720 pixels respectively. “3” denotes the Red, Green and Blue (RGB) channels of the video.

Now, let’s try to be a little more specific

• We want to resize each frame to be 720 pixels in width and 480 pixels in height.

• Not all the frames are required. Let’s just capture exactly 9 frames (at equal intervals) from the video.

• And finally, we’ll also visualize the captured frames.

# Import
from mydia import Videos

# Initialize video path
video_path = r'./static/sample_video/bigbuckbunny.mp4'

# Configuring the parameters
# Setting 'target_size' = (720, 480) : this denotes the new width and height of the frames
# Setting 'num_frames' = 9 : to capture exactly 9 frames
# For more detailed information, view the code documentation.
reader = Videos(target_size=(720, 480),
                num_frames=9)

# Call the 'read()' function to get the required video tensor
video = reader.read(video_path)   # a tensor of shape (1, 9, 480, 720, 3)

# Plot the video frames in a grid
reader.plot(video[0])

Great! Now let’s read the same video in gray scale, instead of RGB.

# Import
from mydia import Videos

# Initialize video path
video_path = r'./static/sample_video/bigbuckbunny.mp4'

# Configuring the parameters
# Other parameters are the same as described above.
# The only additional parameter to modify is 'to_gray'
reader = Videos(target_size=(720, 480),
                to_gray=True,
                num_frames=9)

# Call the 'read()' function to get the required video tensor
video = reader.read(video_path)   # a tensor of shape (1, 9, 480, 720, 1)

# Plot the video frames in a grid
reader.plot(video[0])

Note:
The number of channels for a video in gray scale is 1 (indicated by the last value in the tuple).

Installation

• Install Mydia from PyPI (recommended):

  pip install mydia

• Alternatively, install from Github source:

  First, clone the repository.

  git clone https://github.com/MrinalJain17/mydia.git

  Then, build the module

  cd mydia
  python setup.py install

Requirements

Python 3.x (preferably from the Anaconda Distribution)

The program uses Scikit-video, which requires FFmpeg to be installed on the system. To install FFmpeg on your machine -

For Linux users:

$ sudo apt-get update
$ sudo apt-get install libav-tools

For Windows or MAC/OSX users:

Download the required binaries from here. Extract the zip file and add the location of binaries to the PATH variable

Additional Libraries to install:

Several libraries like Numpy, Pillow and Matplotlib, required for the package come pre-installed with the Anaconda distribution for Python. If you are not using the default anaconda distribution, then first install the packages mentioned above and along with their dependencies.

Also, install the following additional packages:

• Scikit-video

pip install sk-video

• tqdm - Required for displaying the progress bar.

pip install tqdm