Today, with the explosion of artificial intelligence, deep learning is widely used in various fields. Deep learning model training is inseparable from a large number of sample libraries. I have shared before[Python crawler] batch crawling of images from web pages & making data sets. Today I will share with you how to use the OpenCV library to process videos. Frames are extracted to increase the number of sample images. It just so happens that I would also like to share how to combine pictures into videos. Of course, video frame extraction can also be applied to many neighborhoods. I used it here to make samples.
1 Video decomposition pictures (frame splitting)
1.1 Main function introduction + code
cv2.VideoCapture()is a function in the OpenCV library for reading video files or live video streams. It returns a video capture object through which the video can be read, manipulated and released.
Usecv2.VideoCapture() to read every frame in a video file or live video stream. By reading the frames in a loop, you can get all the frames in the video.
The target_frame in the input parameter here refers to the number of frames to save a picture. If you enter 1, all will be saved. If the video frame rate is 60 (60 pictures per second) and you set target_frame to 120, one picture will be saved every two seconds.
def Frame_video(video_path, out_path, target_frame=1):
"""
:param video_path: 需要拆帧的视频路径
:param out_path: 拆帧后图片保存路径
:param target_frame: 抽取帧数间隔，默认为1，即1帧保存1张图片
:return: None
"""
print("-------------------------视频抽帧-------------------------")
if not os.path.exists(out_path):
# 判断文件夹是否存在
os.makedirs(out_path)
video = cv2.VideoCapture() # 初始化一个OpenCV的视频读取对象
video.open(video_path)
count = 0 # 记录当前帧数
image_index = 1000001 # 用于保存图片名称
frames = int(video.get(cv2.CAP_PROP_FRAME_COUNT)) # 获取帧数
print('视频共%s帧，抽取%s帧......' % (frames, int(frames/target_frame)))
while True:
_, frame = video.read()
if frame is None:
# print("第%s帧图片无法打开！" % count)
break
if count % target_frame == 0:
if int((image_index-1000000) / int(frames/target_frame) * 100) in [20, 40, 60, 80]:
print("已提取百分之%s,剩余%s帧......" %
(int((image_index-1000000) / int(frames/target_frame) * 100),
int(frames / target_frame) - image_index + 1000000))
save_path = out_path + "%s.png" % image_index
cv2.imwrite(save_path, frame)
image_index += 1
count += 1
video.release()
print("视频已全部抽帧完成......")
print("-------------------------抽帧完成-------------------------")1.2 Complete code
# -*- coding: utf-8 -*-
"""
@Time ： 2023/10/25 14:26
@Auth ： RS迷途小书童
@File ：Video Frame Images.py
@IDE ：PyCharm
@Purpose：视频拆帧成图片
"""
import os
import sys
import cv2
def Frame_video(video_path, out_path, target_frame=1):
"""
:param video_path: 需要拆帧的视频路径
:param out_path: 拆帧后图片保存路径
:param target_frame: 抽取帧数间隔，默认为1，即1帧保存1张图片
:return: None
"""
print("-------------------------视频抽帧-------------------------")
if not os.path.exists(out_path):
# 判断文件夹是否存在
os.makedirs(out_path)
video = cv2.VideoCapture() # 初始化一个OpenCV的视频读取对象
video.open(video_path)
count = 0 # 记录当前帧数
image_index = 1000001 # 用于保存图片名称
frames = int(video.get(cv2.CAP_PROP_FRAME_COUNT)) # 获取帧数
print('视频共%s帧，抽取%s帧......' % (frames, int(frames/target_frame)))
while True:
_, frame = video.read()
if frame is None:
# print("第%s帧图片无法打开！" % count)
break
if count % target_frame == 0:
if int((image_index-1000000) / int(frames/target_frame) * 100) in [20, 40, 60, 80]:
print("已提取百分之%s,剩余%s帧......" %
(int((image_index-1000000) / int(frames/target_frame) * 100),
int(frames / target_frame) - image_index + 1000000))
save_path = out_path + "%s.png" % image_index
cv2.imwrite(save_path, frame)
image_index += 1
count += 1
video.release()
print("视频已全部抽帧完成......")
print("-------------------------抽帧完成-------------------------")
if __name__ == '__main__':
print("\n-------------------------基础信息-------------------------")
Video_path = r'G:\D.MP4'
save_dir = r'B:\YOLO\18/'
video1 = cv2.VideoCapture() # 初始化一个OpenCV的视频读取对象
if not video1.open(Video_path):
print("无法打开视频，请检查数据！")
sys.exit()
fps = video1.get(cv2.CAP_PROP_FPS) # 获取帧率
frame_count = int(video1.get(cv2.CAP_PROP_FRAME_COUNT)) # 获取视频的总帧数
video1.release() # 清理缓存
duration = frame_count / fps # 计算视频的时长（秒）
print("视频时长为: %ss" % int(duration))
print("视频帧率为: %sFPS" % int(fps))
print("视频帧数为: %s" % int(frame_count))
Frame = int(input("请输入抽取帧数间隔："))
Frame_video(Video_path, save_dir, Frame)
2 picture combination video (framing)
2.1 Main function introduction + code
The cv2.VideoWriter function is used to save the recorded video into a file. It requires specifying parameters such as file path, encoder, frame rate, and video size.
def Image_Frame(images_path, out_path, fps):
"""
:param images_path: 输入需要组帧的图片文件夹路径
:param out_path: 输出视频路径
:param fps: 视频帧率
:return: None
"""
print("-------------------------图片组帧-------------------------")
images_lists = os.listdir(images_path) # images_lists.sort()
image_size = Image.open(os.path.join(images_path, images_lists[0])).size
fourcc = cv2.VideoWriter_fourcc(*"mp4v")
video_writer = cv2.VideoWriter(out_path, fourcc, fps, image_size)
for image_list in images_lists:
image_path = os.path.join(images_path, image_list)
frame = cv2.imread(image_path)
video_writer.write(frame)
print("正在添加:", image_list)
video_writer.release()
print("-------------------------组帧完成-------------------------")2.2 Complete code
# -*- coding: utf-8 -*-
"""
@Time ： 2023/10/25 16:00
@Auth ： RS迷途小书童
@File ：Images Frame Video.py
@IDE ：PyCharm
@Purpose：图片组帧成视频
"""
import os
import cv2
from PIL import Image
def Image_Frame(images_path, out_path, fps):
"""
:param images_path: 输入需要组帧的图片文件夹路径
:param out_path: 输出视频路径
:param fps: 视频帧率
:return: None
"""
print("-------------------------图片组帧-------------------------")
images_lists = os.listdir(images_path) # images_lists.sort()
image_size = Image.open(os.path.join(images_path, images_lists[0])).size
fourcc = cv2.VideoWriter_fourcc(*"mp4v")
video_writer = cv2.VideoWriter(out_path, fourcc, fps, image_size)
for image_list in images_lists:
image_path = os.path.join(images_path, image_list)
frame = cv2.imread(image_path)
video_writer.write(frame)
print("正在添加:", image_list)
video_writer.release()
print("-------------------------组帧完成-------------------------")
if __name__ == "__main__":
Image_path = r'G:\1/'
Out_path = r'G:\1.mp4'
FPS = int(input("请输入帧率："))
Image_Frame(Image_path, Out_path, FPS)
This blog post is shared here. If you have any questions about RS, GIS, and Python, you are welcome to leave a message for communication. Let's learn and progress together!