MGT-python

NEW FUNCTIONALITIES

• _blurfaces.py: Embed audio in video and process files using FFmpeg
• _input_test.py: Add a function to check if FFmpeg is properly installed
• Add github actions workflows and automate documentation on pushes
• Update MusicalGesturesToolbox notebook

ISSUES

• Add VMAF motion score #314
• Fix missing amplitude values in color-waveform #309
• Fix blur_faces() error 2 #308
• Fix progress bar problem blur_faces() #307
• Fix blur_faces numpy issue #306
• Fix mac ffmpeg issues #304
• Fix not compatible with python 3.12 #303
• Fix read/write original files as arrays #294
• Add a completely memory-based processing flow #294
• Fix integrating Github CI, and more #291
• Fix add coloured waveforms #159

NEW FUNCTIONALITIES

• _blurfaces.py: Embed audio in video and process files using FFmpeg
• _input_test.py: Add a function to check if FFmpeg is properly installed

ISSUES

• Add VMAF motion score #314
• Fix missing amplitude values in color-waveform #309
• Fix blur_faces() error 2 #308
• Fix progress bar problem blur_faces() #307
• Fix blur_faces numpy issue #306
• Fix mac ffmpeg issues #304
• Fix not compatible with python 3.12 #303
• Fix read/write original files as arrays #294
• Add a completely memory-based processing flow #294
• Fix integrating Github CI, and more #291
• Fix add coloured waveforms #159

NEW FUNCTIONALITIES

• Add blend.py

ISSUES

• Fix keyframe history images #203
• Fix problems downloading models for pose #299
• Fix index out of range in audio object #300
• Fix harmonic/percussive waveform #301

NEW FUNCTIONALITIES

• Add harmonic/percussive source separation #272
• Add optional title SSM #286
• Add MgAudio object #273

ISSUES

• Fix motion() chokes on rotated video files #289
• Fix crop preview #288
• Fix metadata rotation #287
• Fix titles on waveforms and spectrograms #285
• Fix SSM on videograms #284
• Fix time in trimmed videos #279
• Fix Error with SSM #278

ISSUES

Fix video playback in Colab #277
Fix MP4 conversion #282
Normalized pixel values in motion() #290

NEW FUNCTIONALITIES

• Update MGT promo picture

ISSUES

• Fix filtering functions don't work #274
• Fix strange inversion #275
• Fix SSM issue #276

NEW FUNCTIONALITIES

• Create a new filter_frame_ffmpeg function using ffmpeg
• Create a new transform_frame function for transforming the bytes into numpy arrays
• Add more motion analysis #210 with dense optical flow velocity, acceleration and entropy
• Add background subtraction #270 based on a static image

ISSUES

• Reimplement motion() with ffmpeg #109

NEW FUNCTIONALITIES

• Add self-similarity matrix #173 (SSM) for motiongrams, spectrogram, chromagram and tempogram features
• Add analysis examples #211 and add violin plot to QoM
• Add heatmap of face detection paramater #262 to the blur_faces function in order to create some visualizations based on the faces coordinates
• Add fixed-frames in addition to the skip frames function #197 in order to specify a fixed number of frames (or the keyframes with frames=-1) to extract from the video
• Add audio plots without lines and units #266 in order to plot raw audio waveform and spectrograms

ISSUES

• Fix slow skip frames #261 by hardcoding .avi files to the skip function
• Fix issues with info() in ipython #267
• Fix unit parameter for audio_descriptors in motionplots()

Fix issues, add and update several functions:

• Fix bug in impact_detections() function

• Add Area of motion (AoM) in the motiondata() and motionplots() functions

• Add audio_descriptors parameter to the motionplots() function

• Add info() function to retrieve technical metadata from video, audio and format

• Update nth_frames parameter in grid() function

• Update data storage in blur_faces() function

• Update waveform() and spectrogram() functions and add a raw plotting parameter

Change documentation folder to docs, fix issues and update several functions:

• Fix issue with manual cropping in _cropvideo.py
• Add parameters to save the scaled coordinates of the face mask in different file formats (csv, tsv, and txt) in _blurfaces.py
• Add new function grid.py to generate frame strip video preview using ffmpeg

This new release added the function blur_faces useful for an automatic anonymization of faces in videos. Updated documentation.

This new version added the warping_audiovisual_beats functions as well as the directograms and impacts functions.
Breaking change: we also changed the name of the MgObject to the MgVideo.

Lots of bug-fixes and improvements, several new functions, tests, and better documentation.

First Release of the Musicial Gestures Toolbox for Python 3.