短视频封面抽取和标题自动化生成

思路

1.对人像直播视频抽取关键帧

2.对关键帧做质量打分排序

3.利用语音和视频字幕抽取内容文本

4.利用文本做文章标题自动生成

5.基于模版合成视频封面

人像抽取部分

0.视频字幕去除

# 安装 AgentOCR
!pip install agentocr

# 安装 CPU 版本 ONNXRuntime
!pip install onnxruntime
import cv2
import numpy as np
from PIL import Image
import matplotlib.pyplot as plt

def image_remove_word(img_path = 'images/frame.177000000.jpg'):
    mask_threth = 50

    # 设置测试图片路径
    img_path = img_path
    img = cv2.imread(img_path) #自己qq截图一张图片就行，要大于下面的坐标点

    # 通过 config 参数来进行模型配置，内置多国语言的配置文件
    ocr = OCRSystem(config='ch')

    # 调用 OCR API 进行全流程识别
    result = ocr.ocr(img_path)

    coordinates = []
    # 打印结果
    for line in result:
        coordinate1 = []
        a0 = line[0][0][0]-5
        a1 = line[0][0][1]-5
        b0 = line[0][1][0]+5
        c1 = line[0][2][1]+5
        coordinate1.append([[a0,a1],[b0,a1],[b0,c1],[a0,c1]])
        coordinate1 = np.array(coordinate1, np.int32)
        coordinates.append(coordinate1)
        #print(line[0])

    # binary mask
    mask = np.zeros(img.shape[:2], dtype=np.int8)
    mask = cv2.fillPoly(mask, coordinates,255)

    cv2.imwrite('images/mask1.png', mask)
    mask = cv2.imread('images/mask1.png',0)

    dst = cv2.inpaint(img,mask,3,cv2.INPAINT_TELEA)
    cv2.imwrite(img_path,dst)

#img_path = 'images/frame.177000000.jpg'
#image_remove_word(img_path)

def getFrame(video_name, save_path):#将视频逐帧保存为图片

    video = cv2.VideoCapture(video_name)

    # 获取视频帧率
    fps = video.get(cv2.CAP_PROP_FPS)
    print(fps)
    
    # 获取画面大小
    width = int(video.get(cv2.CAP_PROP_FRAME_WIDTH))
    height = int(video.get(cv2.CAP_PROP_FRAME_HEIGHT))
    size = (width, height)

    # 获取帧数
    frame_num = str(video.get(7))
    name = int(math.pow(10, len(frame_num)))
    ret, frame = video.read()
    #ocr = OCRSystem(config='ch')
    while ret:
        cv2.imwrite(save_path + str(name) + '.jpg', frame)
        #imagepath = save_path + str(name) + '.jpg'
        #image_remove_word(imagepath,ocr)
        ret, frame = video.read()
        name += 1
    video.release()
    return fps, size, frame_num

def writeVideo(humanseg, fps, size):# 写入视频
    
    fourcc = cv2.VideoWriter_fourcc(*'mp4v')
    out = cv2.VideoWriter('./movies/green.mp4', fourcc, fps, size)

    # 为每一帧设置背景
    files = os.listdir(humanseg)
    #将文件名按编号顺序排序
    file_num_sort = []
    for file in files:
        file_num_sort.append(int(file.split('.')[0]))
    file_num_sort.sort()
    #print(file_num_sort)
    file_sort = []
    for file_num in file_num_sort:
        file_sort.append('./frames/'+str(file_num)+'.jpg')
    #print(file_sort)

    ocr = OCRSystem(config='ch')
    for file in file_sort:
        print('将'+file+'写入视频')
        #imagepath = save_path + str(name) + '.jpg'
        image_remove_word(file,ocr)
        im_array = cv.imread(file)#setImageBg(file, bg_im)
        out.write(im_array)
    out.release()
  
def getMusic(video_name):#获取音频

    video = VideoFileClip(video_name)
    return video.audio

def addMusic(video_name, audio):#给video_name添加音频
    
    video = VideoFileClip(video_name)#读取
    video = video.set_audio(audio)#设置
    video.write_videofile(output_video)#保存
    
!pip install moviepy
import os
import math
#将视频按帧保存为图片
frame_path = "./frames/"
if not os.path.exists(frame_path):
    os.makedirs(frame_path)
video_name = './movies/whitebottle.mp4'
fps, size, frame_number = getFrame(video_name, frame_path)
print(fps, size, frame_number)

from moviepy.editor import *
import cv2 as cv
# 最终视频的保存路径
output_video = './movies/result.mp4'
humanseg_path = 'frames'
writeVideo(humanseg_path, fps, size)
addMusic('./movies/green.mp4', getMusic(video_name))

#https://aistudio.baidu.com/aistudio/projectdetail/3919465?forkThirdPart=1

1.视频背景去除

!pip install --quiet av pims
import torch

model = torch.hub.load("PeterL1n/RobustVideoMatting", "mobilenetv3").cuda() # or "resnet50"
convert_video = torch.hub.load("PeterL1n/RobustVideoMatting", "converter")

convert_video(
    model,                           # The loaded model, can be on any device (cpu or cuda).
    input_source='input.mp4',        # A video file or an image sequence directory.
    downsample_ratio=None,           # [Optional] If None, make downsampled max size be 512px.
    output_type='video',             # Choose "video" or "png_sequence"
    output_composition='com.mp4',    # File path if video; directory path if png sequence.
    output_alpha="pha.mp4",          # [Optional] Output the raw alpha prediction.
    output_foreground="fgr.mp4",     # [Optional] Output the raw foreground prediction.
    output_video_mbps=4,             # Output video mbps. Not needed for png sequence.
    seq_chunk=12,                    # Process n frames at once for better parallelism.
    num_workers=1,                   # Only for image sequence input. Reader threads.
    progress=True                    # Print conversion progress.
)

2.视频关键帧提取

!pip install av
import av
import os
import shutil

# 视频提取关键帧工具类（支持批量视频）
class PyAvUtils:
    def __init__(self, video_dir, keyframe_dir):
        self.video_dir = video_dir
        self.keyframe_dir = keyframe_dir
 
    # 提取关键帧并保存
    def do_video2image(self):
        fo