将字符串转换回numpy数组 [英] Convert string back to numpy array

问题描述

我正在使用python进行视频聊天.我正在使用CV2捕获当前图像(这给了我一个numpy数组)，然后将numpy发送到我的服务器.现在我在服务器上有一个字符串，我需要将其解码回一个numpy数组.

I'am working on a video chat in python. I'am using CV2 to capture the current image(this gives me a numpy array), then I send the numpy over to my server. Now I have a string at the server and i need to decode it back to a numpy array.

我正在使用Python 3.7，但还没有想到.

I'am using Python 3.7 and i dindn't come up with somethink yet.

    #client 
    #capture 500 frames
    while i < 500:
        i = i + 1
        # Capture frame-by-frame
        ret, frame = cap.read()
        #send data                           
        client_socket.send(bytes(str(ret) + "###" + str(frame), "utf-8"))


    #server
    #split ret and frame
    ret, frame = str(conn.recv(16000)).split("###")
    gray = cv2.cvtColor(frame.toNumpyArray #PseudoMethod  <-- Here
    ,cv2.COLOR_BGR2GRAY)

我只需要一种将字符串转换回numpy数组的方法. 如果我将其打印出来，则字符串看起来像这样:

I only need a method to convert the string back to a numpy array. If i print it out,the string looks like this:

b'[[[[128255255]] \ n [125255255] \ n [107255255] \ n ... \ n [102130167] \ n [102128172] \ n [102128 172]] \ n \ n [[128 255 255] \ n [127255255] \ n [108255255] \ n ... \ n [102130167] \ n [102128172] \ n [102 128 172]] \ n \ n [[111 255 255] \ n [111 255 255] \ n [109255255] \ n ... \ n [99131169] \ n [99131169] \ n [ 99 131 169]] \ n \ n ... \ n \ n [[27 64 95] \ n [29 67 97] \ n [24 66 98] \ n ... \ n [73117160] \ n [70 119 161] \ n [70 119 161]] \ n \ n [[18 71 81] \ n [20 74 83] \ n [30 67 93] \ n ... \ n [77117159] \ n [74 118 163] \ n [74 118 163]] \ n \ n [[14 68 77] \ n [19 73 82] \ n [30 67 93] \ n ... \ n [77117159] \ n [74 118 163] \ n [74 118 163]]]'

b'[[[128 255 255]\n [125 255 255]\n [107 255 255]\n ...\n [102 130 167]\n [102 128 172]\n [102 128 172]]\n\n [[128 255 255]\n [127 255 255]\n [108 255 255]\n ...\n [102 130 167]\n [102 128 172]\n [102 128 172]]\n\n [[111 255 255]\n [111 255 255]\n [109 255 255]\n ...\n [ 99 131 169]\n [ 99 131 169]\n [ 99 131 169]]\n\n ...\n\n [[ 27 64 95]\n [ 29 67 97]\n [ 24 66 98]\n ...\n [ 73 117 160]\n [ 70 119 161]\n [ 70 119 161]]\n\n [[ 18 71 81]\n [ 20 74 83]\n [ 30 67 93]\n ...\n [ 77 117 159]\n [ 74 118 163]\n [ 74 118 163]]\n\n [[ 14 68 77]\n [ 19 73 82]\n [ 30 67 93]\n ...\n [ 77 117 159]\n [ 74 118 163]\n [ 74 118 163]]]'

对不起，我的英语不好，我是德国学生.

Sorry for my bad english, I'am a german student.

推荐答案

以下一对程序演示了一种通过网络套接字通信numpy ndarray对象的方法.客户端使用save方法将数组转换为字节流，将流写入BytesIO对象，然后通过套接字将其发送到服务器:

The following pair of programs demonstrates one way to commmunicate a numpy ndarray object across a network socket. The client converts the array to a byte stream using the save method, writing the stream to a BytesIO object that is then sent across the socket to the server:

import numpy as np
import socket
from io import BytesIO

# Create an output socket connected to server
sout = socket.create_connection(('127.0.0.1', 6543))

# Create data and write binary stream out to socket

a = np.array([[1.1, 2.2, 3.3],
              [4.4, 5.5, 6.6],
              [7.7, 8.8, 9.9]])

b = BytesIO()
np.save(b, a)

sout.send(b.getvalue())
sout.close()

服务器在适当的网络地址上侦听，接收数据，直到结束套接字关闭为止.然后，它将接收到的数据转换为BytesIO对象，numpy的load函数从该对象恢复结构:

The server listens on the appropriate network address, receiving data until the ending socket closes. It then converts the received data into a BytesIO object, from which numpy's load function recovers the structure:

import numpy as np
import socket
from io import BytesIO

# Create socket listening on correct port
s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
s.bind(('127.0.0.1', 6543))
s.listen()

# Accept a connection and receive data
sock, address = s.accept()
data = b''
while True:
    indt = sock.recv(1024)
    if not indt:
        break
    data += indt

# Take data and recast to np.ndarray
data = BytesIO(data)

b = np.load(data)
print(type(b), b, sep='\n')

运行服务器的输出如下:

The output from running the server is as follows:

<class 'numpy.ndarray'>
[[1.1 2.2 3.3]
 [4.4 5.5 6.6]
 [7.7 8.8 9.9]]

可以通过多种方式来优化此代码，但这应该给您足够的动力.

There are various ways in which this code could be optimised, but this should give you enough to get going.

这篇关于将字符串转换回numpy数组的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！