Videopypeline

A fancy wrapper for opencv inspired by the keras model API.

Install

pip install videopypeline

Usage

videopypeline can be used to process any datatype. Nonetheless, its primary use is intended for videos. The examples below show some use cases. Refer to the Python-Notebook in this repository to see a more established usage example.

Linear pipeline

import videopypeline as vpl

video_path = './path/to/video.mp4'
output_path = './path/to/output.mp4'

# Setup pipeline model
raw_video = videopypeline.generators.ReadVideoFile(video_path)
grey = videopypeline.functions.Rgb2Greyscale()(raw_video)
crop = videopypeline.functions.Crop((100, 200), (500, 500))(grey)
smooth1 = videopypeline.functions.Smooth(101)(crop)
stats = videopypeline.core.Action(lambda frame: print(frame.mean(), frame.std()))(smooth1)
writer = videopypeline.actions.VideoWriter(output_path, 30, aggregate=True, collect=False, verbose=True)(stats)

# Invoke pipeline
_ = writer()

Tree pipeline

import videopypeline as vpl

video1_path = './path/to/video1.mp4'
video2_path = './path/to/video2.mp4'
output_path = './path/to/output.mp4'

# Setup pipeline model
raw_video1 = videopypeline.generators.ReadVideoFile(video1_path)
grey1 = videopypeline.functions.Rgb2Greyscale()(raw_video1)
smooth1 = videopypeline.functions.Smooth(101)(grey1)

raw_video2 = videopypeline.generators.ReadVideoFile(video2_path)
grey2 = videopypeline.functions.Rgb2Greyscale()(raw_video2)
smooth2 = videopypeline.functions.Smooth(101)(grey2)

# This node has two parents
diff = videopypeline.core.Function(lambda *frames: frames[0] - frames[1])([smooth1, smooth2])
writer = videopypeline.actions.VideoWriter(output_path, 30, aggregate=True, collect=False, verbose=True)(diff)

# Invoke pipeline
_ = writer()

Graph pipeline

</code></pre>
<h3>Filter pipeline</h3>
<pre lang="py"><code>

Terminology

Node

Generator

Function

Remarks

When adding custom Nodes, which inherit from vpl.core.Function, make sure the output object is different from the input object, as this could lead to unexpected behaviour.

Action

Pipeline