python - 我可以使用所有 CPU 和内核来处理图像吗？

Question

我是Python编码新手

所以我有一个进行人脸识别的项目，我有20个文件夹，每个文件夹包含4张图像。但是当我运行代码时，编码过程花费了很多时间。

这是我的代码:

import face_recognition
import cv2
import numpy as np
import os 
import glob

path = './dataset'
folders = [f for f in glob.glob(path + '**/*', recursive=True)]

known_face_encodings = []
known_face_names = []

for f in folders:
    names = f.split('/')[2]
    print('encoding file : {}'.format(names))
    for images_f in glob.glob(f + '**/*.jpg'):

        images = face_recognition.load_image_file(images_f)
        location = face_recognition.face_locations(images)
        images_encoding = face_recognition.face_encodings(images, known_face_locations = location)[0]
        known_face_encodings.append(images_encoding)
        known_face_names.append(names)

我可以做些什么来加快编码过程吗？如何？提前致谢

Answer 1

您可以使用它来将处理分成不同的线程:

import threading

def face_recognition_function( f ) :
    names = f.split('/')[2]
    print('encoding file : {}'.format(names))
    for images_f in glob.glob(f + '**/*.jpg'):

        images = face_recognition.load_image_file(images_f)
        location = face_recognition.face_locations(images)
        images_encoding = face_recognition.face_encodings(images, known_face_locations = location)[0]
        known_face_encodings.append(images_encoding)
        known_face_names.append(names)

threads = []
for f in folders:
    t = threading.Thread(target=face_recognition_function, args=(f,))
    t.start()
    threads.append( t )

# threads work now on your images

for t in threads :
    t.join()  # wait for the completion

当前版本的代码为每个文件夹启动一个新线程。如果这以某种方式使您的计算机耗尽内存或导致不可预测的结果，您可能需要创建较少的并行线程。不过，对于 20 个文件夹，我认为这里不会发生任何问题。