本文对常见的图像读取函数进行了速度测试

1. jpeg4py 第三方库
2. Pillow
3. opencv-python

结论：

单线程下，jpeg4py速度是pillow的3倍，是cv的2.35倍

import jpeg4py
import numpy as np
from PIL import Image
from  matplotlib import pyplot as plt
from cv2 import imread

path = '/home/ryf_stu01/pytracking/dataset/got10k/GOT-10k_Train_004502/00000015.jpg'

1. jpeg4py

%timeit img = jpeg4py.JPEG(path).decode()

5.18 ms ± 51.8 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)

2. Pillow(PIL)

%timeit np.array(Image.open(path))

25.3 ms ± 447 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)

3. opencv

%timeit imread(path)

12.3 ms ± 82 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)