La vida es demasiado corta, uso Python
Preámbulo
Cuando miramos videos, a veces hay algunos mosaicos extraños que afectan nuestra experiencia de visualización, entonces, ¿cómo se agregan exactamente estos mosaicos?
¡Esta vez usaremos Python para codificar videos automáticamente!
Listo para trabajar
Todavía podemos usar Python3.8 y pycharm2021 para el medio ambiente
Principio de implementación
- Divide el video en audio e imagen;
- El rostro aparece en la pantalla y se compara el objetivo, y se codifica el rostro correspondiente;
- Agregue sonido al video procesado;
módulo
Instale manualmente el módulo cv2, pip install opencv-python installation
Si se informa un error durante la instalación, no se instalará Consulte el artículo en la parte superior de mi página de inicio.
herramienta material
Necesitamos instalar la herramienta de transcodificación de audio y video ffmpeg
Todos los materiales se pueden obtener escaneando el código de la izquierda
análisis de código
Importar módulos requeridos
import cv2
import face_recognition # 人脸识别库 99.7% cmake dlib face_recognition
import subprocess
Convertir video a audio
def video2mp3(file_name):
"""
:param file_name: 视频文件路径
:return:
"""
outfile_name = file_name.split('.')[0] + '.mp3'
cmd = 'ffmpeg -i ' + file_name + ' -f mp3 ' + outfile_name
print(cmd)
subprocess.call(cmd, shell=False)
añadir mosaico
def mask_video(input_video, output_video, mask_path='mask.jpg'):
"""
:param input_video: 需打码的视频
:param output_video: 打码后的视频
:param mask_path: 打码图片
:return:
"""
# 读取图片
mask = cv2.imread(mask_path)
# 读取视频
cap = cv2.VideoCapture(input_video)
# 视频 fps width height
v_fps = cap.get(5)
v_width = cap.get(3)
v_height = cap.get(4)
# 设置写入视频参数 格式MP4
# 画面大小
size = (int(v_width), int(v_height))
fourcc = cv2.VideoWriter_fourcc('m', 'p', '4', 'v')
# 输出视频
out = cv2.VideoWriter(output_video, fourcc, v_fps, size)
# 已知人脸
known_image = face_recognition.load_image_file('tmr.jpg')
biden_encoding = face_recognition.face_encodings(known_image)[0]
cap = cv2.VideoCapture(input_video)
while (cap.isOpened()):
ret, frame = cap.read()
if ret:
# 检测人脸
# 人脸区域
face_locations = face_recognition.face_locations(frame)
for (top_right_y, top_right_x, left_bottom_y, left_bottom_x) in face_locations:
print((top_right_y, top_right_x, left_bottom_y, left_bottom_x))
unknown_image = frame[top_right_y - 50:left_bottom_y + 50, left_bottom_x - 50:top_right_x + 50]
if face_recognition.face_encodings(unknown_image) != []:
unknown_encoding = face_recognition.face_encodings(unknown_image)[0]
# 对比人脸
results = face_recognition.compare_faces([biden_encoding], unknown_encoding)
# [True]
# 贴图
if results == [True]:
mask = cv2.resize(mask, (top_right_x - left_bottom_x, left_bottom_y - top_right_y))
frame[top_right_y:left_bottom_y, left_bottom_x:top_right_x] = mask
out.write(frame)
else:
break
Agregar audio a la imagen
def video_add_mp3(file_name, mp3_file):
"""
:param file_name: 视频画面文件
:param mp3_file: 视频音频文件
:return:
"""
outfile_name = file_name.split('.')[0] + '-f.mp4'
subprocess.call('ffmpeg -i ' + file_name + ' -i ' + mp3_file + ' -strict -2 -f mp4 ' + outfile_name, shell=False)
código completo
import cv2
import face_recognition # 人脸识别库 99.7% cmake dlib face_recognition
import subprocess
def video2mp3(file_name):
outfile_name = file_name.split('.')[0] + '.mp3'
cmd = 'ffmpeg -i ' + file_name + ' -f mp3 ' + outfile_name
print(cmd)
subprocess.call(cmd, shell=False)
def mask_video(input_video, output_video, mask_path='mask.jpg'):
# 读取图片
mask = cv2.imread(mask_path)
# 读取视频
cap = cv2.VideoCapture(input_video)
# 视频 fps width height
v_fps = cap.get(5)
v_width = cap.get(3)
v_height = cap.get(4)
# 设置写入视频参数 格式MP4
# 画面大小
size = (int(v_width), int(v_height))
fourcc = cv2.VideoWriter_fourcc('m', 'p', '4', 'v')
# 输出视频
out = cv2.VideoWriter(output_video, fourcc, v_fps, size)
# 已知人脸
known_image = face_recognition.load_image_file('tmr.jpg')
biden_encoding = face_recognition.face_encodings(known_image)[0]
cap = cv2.VideoCapture(input_video)
while (cap.isOpened()):
ret, frame = cap.read()
if ret:
# 检测人脸
# 人脸区域
face_locations = face_recognition.face_locations(frame)
for (top_right_y, top_right_x, left_bottom_y, left_bottom_x) in face_locations:
print((top_right_y, top_right_x, left_bottom_y, left_bottom_x))
unknown_image = frame[top_right_y - 50:left_bottom_y + 50, left_bottom_x - 50:top_right_x + 50]
if face_recognition.face_encodings(unknown_image) != []:
unknown_encoding = face_recognition.face_encodings(unknown_image)[0]
# 对比人脸
results = face_recognition.compare_faces([biden_encoding], unknown_encoding)
# [True]
# 贴图
if results == [True]:
mask = cv2.resize(mask, (top_right_x - left_bottom_x, left_bottom_y - top_right_y))
frame[top_right_y:left_bottom_y, left_bottom_x:top_right_x] = mask
out.write(frame)
else:
break
def video_add_mp3(file_name, mp3_file):
outfile_name = file_name.split('.')[0] + '-f.mp4'
subprocess.call('ffmpeg -i ' + file_name + ' -i ' + mp3_file + ' -strict -2 -f mp4 ' + outfile_name, shell=False)
if __name__ == '__main__':
# 1.
video2mp3('cut.mp4')
# 2.
mask_video(input_video='cut.mp4',output_video='output.mp4')
# 3.
video_add_mp3(file_name='output.mp4',mp3_file='cut.mp3')
¡Hermanos, pruébenlo!
Bienvenido a discutir e intercambiar en el área de comentarios ~