DeepFrame

Python module to extract frames from a video file.

Installation

pip install deepframe

API

Package provides a simple API to get video frames from a video file. The frames are returned as buffer view objects which can be converted to numpy arrays.

import deepframe
import numpy as np

# Extract whole video as a buffer view
buffer_view = deepframe.extract_frames('path/to/video.mp4')
frames = np.asarray(buffer_view)

import deepframe
import numpy as np
# Extract first 100 frames from a video file
buffer_view = deepframe.extract_frames('path/to/video.mp4', slice(0, 100))
frames = np.asarray(buffer_view)

import deepframe
import numpy as np
# Extract each 10th frame from a video file
buffer_view = deepframe.extract_frames('path/to/video.mp4', slice(0, None, 10))
frames = np.asarray(buffer_view)

C++ Build

Requirements

• vcpkg

You also need to install python requirements. pip install -r requirements-dev.txt

VCPkg Setup

VCPkg is installed via a git repository which a VCPKG_ROOT environment variable points to. The VCPKG_ROOT can be whatever location on your machine you want.

git clone https://github.com/microsoft/vcpkg.git

cd vcpkg
export VCPKG_ROOT=`pwd`

./bootstrap-vcpkg.sh

echo "export VCPKG_ROOT=$VCPKG_ROOT" >> ~/.zshrc # or path to your shell config file
echo "export PATH=$PATH:$VCPKG_ROOT" >> ~/.zshrc # or path to your shell config file

Build

Debug and Release modes are supported. From project root run:

python scripts/build.py debug
python scripts/build.py release

After the build you can find the python package in ./build/{mode} or the latest one is always available in ./deepframe/ which is used to run the tests.

Test

For testing we use pytest. To run all tests simply call from project root.

pytest

Build wheel

From project root call

python -m build

Once the run finishes you will find the files under ./dist/ directory.