Und wenn ja wie das geht
Ich hatte mir die Doku durchgelesen bin allerdings nicht so schlau daraus geworden für meine Zwecke.
In Anaconda und für ein einzelnes Skript ist das ja kein Problem (Doku) aber geht das auch für einzelne Methoden bzw. eine einzelne Klasse in einem größeren Projekt ?
Hintergrund: das Projekt wird mit fbs gestartet und auch gefreezt
Hier die Klasse:
Code: Alles auswählen
import os
import skvideo.io
import cv2
from pathlib import Path
from moviepy.editor import AudioFileClip
from model.data import VisualizerVideo, BoardVideo, Audio, SlideVideo, SpeakerVideo
class VideoSplitter:
"""
This class handles the video and audio splitting
"""
def __init__(self, folder_path, folder_name, video_data):
"""
Constructor of the class
@param folder_path: the path to the project folder
@param folder_name: the name of the project folder
@param video_data: the path of the video file
"""
self.folder_path = folder_path
self.folder_name = folder_name
self.video_data = video_data
self.files = []
self.audio_files = []
self.frame = 0
self.number_frames = 0
def cut_video(self, update_progress):
"""
a method which cut the modivideos from the "main" video
@param update_progress: a function which handles the progressbar countprocess
"""
self.frame = 0
reader = skvideo.io.FFmpegReader(self.video_data, {}, {})
videometadata = skvideo.io.ffprobe(self.video_data)
self.frame_rate = videometadata['video']['@avg_frame_rate']
self.number_frames = int(videometadata['video']['@nb_frames'])
folder = Path(self.folder_path, self.folder_name)
board_filename = os.path.join(folder, 'board.mp4')
slide_filename = os.path.join(folder, 'slides.mp4')
visualizer_filename = os.path.join(folder, 'visualizer.mp4')
board_out = skvideo.io.FFmpegWriter(board_filename, inputdict={
"-r": self.frame_rate
})
slide_out = skvideo.io.FFmpegWriter(slide_filename, inputdict={
"-r": self.frame_rate
})
visualizer_out = skvideo.io.FFmpegWriter(visualizer_filename, inputdict={
"-r": self.frame_rate
})
# iterate through the frames
for frame in reader.nextFrame():
board_out.writeFrame(frame[183:537, 11:637])
slide_out.writeFrame(frame[183:537, 720:1187])
visualizer_out.writeFrame(frame[183:537, 640:1280])
self.frame += 1
if self.frame % 30 == 0:
update_progress((int)(self.frame/self.number_frames*100))
board_out.close()
slide_out.close()
visualizer_out.close()
self.files.append(board_filename)
self.files.append(slide_filename)
self.files.append(visualizer_filename)
self.__board_video = BoardVideo(board_filename)
self.__slide_video = SlideVideo(slide_filename)
self.__visualizer_video = VisualizerVideo(visualizer_filename)
def get_board_video(self):
return self.__board_video
def get_slide_video(self):
return self.__slide_video
def get_visualizer_video(self):
return self.__visualizer_video
def get_speaker_video(self):
return self.__speaker_video
def cut_audio_from_video(self):
"""
a method to get the audio from a video and save it in the project folder
and create a object of this
@return: a audio object which contains the path
"""
folder = Path(self.folder_path, self.folder_name)
audio_from_video = 'audio.mp3'
audio = AudioFileClip(self.video_data)
audio.write_audiofile(os.path.join(folder, audio_from_video), verbose=False, logger=None)
extracted_audio = Path(folder, audio_from_video)
self.audio_files.append(extracted_audio)
return Audio(extracted_audio)
def cut_zoom_video(self):
"""
a method which track the speaker in the video and cut a video from this
"""
folder = Path(self.folder_path, self.folder_name)
speaker_filename = os.path.join(folder, 'speaker.mp4')
video = self.get_board_video()
cap = cv2.VideoCapture(str(video))
#fgbg = cv2.createBackgroundSubtractorMOG2()
fgbg = cv2.bgsegm.createBackgroundSubtractorMOG()
is_ok, frame = cap.read()
r, h, c, w = 180, 400, 400, 550
track_window = (c,r,w,h)
roi = frame[r:h, c:w]
term_crit = (cv2.TERM_CRITERIA_EPS | cv2.TERM_CRITERIA_COUNT, 5 , 1)
fourcc = cv2.VideoWriter_fourcc(*'mp4v')
out = cv2.VideoWriter(speaker_filename, fourcc , 21, (150, 220))
while True:
is_ok, frame = cap.read()
gmask = fgbg.apply(frame)
if is_ok == True:
is_ok, track_window = cv2.meanShift(gmask, track_window, term_crit)
r, h, c, w = track_window
#rect = cv2.rectangle(frame, track_window, 255, 2)
out.write(track_window)
else:
break
cap.release()
out.release()
self.files.append(speaker_filename)
self.__speaker_video = SpeakerVideo(speaker_filename)
PS: cut_zoom_video wird noch mit skvideo umgeschrieben (durch diesen Fehler)
Danke für eure Hilfe
MfG Felix
