Home  >  Article  >  Backend Development  >  Guide you step by step to implement cutout and change the background color through the Python calling interface

Guide you step by step to implement cutout and change the background color through the Python calling interface

青灯夜游
青灯夜游forward
2022-10-21 20:03:594331browse

Guide you step by step to implement cutout and change the background color through the Python calling interface

Sometimes we need to change the background color of our ID photos, and we don’t have time to go to the photo studio to take pictures. It’s not easy to cut out pictures with PS, so today I will share with you how to use Python to cut out pictures. , and change the background color

1. Register a Baidu AI account and create a portrait segmentation application

  • Baidu portrait segmentation homepage: follow the steps to register and log in , real-name authentication is enough.

  • Find Human Analysis on the console home page

Guide you step by step to implement cutout and change the background color through the Python calling interface

Create application

## You can write whatever you want in Guide you step by step to implement cutout and change the background color through the Python calling interface

#. New users have to get free resources, otherwise they won’t be able to use them.

Guide you step by step to implement cutout and change the background color through the Python calling interface

#After creation is completed, record the values ​​of API Key and Secret Key in the application list, which will be used later.

Guide you step by step to implement cutout and change the background color through the Python calling interface

At this point, the tasks of registering an account and creating an application are completed.

Guide you step by step to implement cutout and change the background color through the Python calling interface

2. Code implementation

1.Introduce the library
import os
import requests
import base64
import cv2
import numpy as np
from PIL import Image
from pathlib import Path

path = os.getcwd()
paths = list(Path(path).glob('*'))

2. Get Access Token
def get_access_token():
    url = 'https://aip.baidubce.com/oauth/2.0/token'
    data = {
        'grant_type': 'client_credentials',  # 固定值
        'client_id': '替换成你的API Key',  # 在开放平台注册后所建应用的API Key
        'client_secret': '替换成你的Secret Key'  # 所建应用的Secret Key
    }
    res = requests.post(url, data=data)
    res = res.json()
    access_token = res['access_token']
    return access_token

Core Code

If you have any questions about the article, you can send me a private message or come here https://jq. qq.com/?_wv=1027&k=s5bZE0K3

def removebg():
    try:
        request_url = "https://aip.baidubce.com/rest/2.0/image-classify/v1/body_seg"
        # 二进制方式打开图片文件
        f = open(name, 'rb')
        img = base64.b64encode(f.read())
        params = {"image":img}
        access_token = get_access_token()
        request_url = request_url + "?access_token=" + access_token
        headers = {'content-type': 'application/x-www-form-urlencoded'}
        response = requests.post(request_url, data=params, headers=headers)
        if response:
            res = response.json()["foreground"]
            png_name=name.split('.')[0]+".png"
            with open(png_name,"wb") as f:
                data = base64.b64decode(res)
                f.write(data)
            fullwhite(png_name) #png图片底色填充,视情况舍去
            png_jpg(png_name) #png格式转jpg,视情况舍去
            os.remove(png_name) #删除原png图片,视情况舍去
            print(name+"\t处理成功!")
    except Exception as e:
        pass

4. Image background color filling
def fullwhite(png_name):
    im = Image.open(png_name)
    x,y = im.size
    try:
        p = Image.new('RGBA', im.size, (255,255,255))        # 使用白色来填充背景,视情况更改
        p.paste(im, (0, 0, x, y), im)
        p.save(png_name)
    except:
        pass

5. Image compression
#compress_rate:数值越小照片越模糊
def resize(compress_rate = 0.5):
    im = Image.open(name)
    w, h = im.size
    im_resize = im.resize((int(w*compress_rate), int(h*compress_rate)))
    resize_w, resieze_h = im_resize.size
    #quality 代表图片质量,值越低越模糊
    im_resize.save(name)
    im.close()

6. Get the image size
def get_size():
    size = os.path.getsize(name)
    return size / 1024

7.png format to jpg
def png_jpg(png_name):
    im = Image.open(png_name)
    bg=Image.new('RGB',im.size,(255,255,255))
    bg.paste(im)
    jpg_name = png_name.split('.')[0]+".jpg"
    #quality 代表图片质量,值越低越模糊
    bg.save(jpg_name,quality=70)
    im.close()

8.Main Function
if __name__ == '__main__':
    for i in paths:
        name = os.path.basename(i.name)
        if(name==os.path.basename(__file__)):
            continue
        size = get_size()
        ##照片压缩
        while size >=900:
            size = get_size()
            resize()   
        removebg()
        print(" ")

9. Complete code

If you have any questions about the article, you can send me a private message or come here https://jq.qq.com /?_wv=1027&k=s5bZE0K3

#人像分割
import os
import requests
import base64
import cv2
import numpy as np
from PIL import Image
from pathlib import Path

path = os.getcwd()
paths = list(Path(path).glob('*'))

def get_access_token():
    url = 'https://aip.baidubce.com/oauth/2.0/token'
    data = {
        'grant_type': 'client_credentials',  # 固定值
        'client_id': '替换成你的API Key',  # 在开放平台注册后所建应用的API Key
        'client_secret': '替换成你的Secret Key'  # 所建应用的Secret Key
    }
    res = requests.post(url, data=data)
    res = res.json()
    access_token = res['access_token']
    return access_token
def png_jpg(png_name):
    im = Image.open(png_name)
    bg=Image.new('RGB',im.size,(255,255,255))
    bg.paste(im)
    jpg_name = png_name.split('.')[0]+".jpg"
    #quality 代表图片质量,值越低越模糊
    bg.save(jpg_name,quality=70)
    im.close()

#compress_rate:数值越小照片越模糊
def resize(compress_rate = 0.5):
    im = Image.open(name)
    w, h = im.size
    im_resize = im.resize((int(w*compress_rate), int(h*compress_rate)))
    resize_w, resieze_h = im_resize.size
    #quality 代表图片质量,值越低越模糊
    im_resize.save(name)
    im.close()
    
def get_size():
    size = os.path.getsize(name)
    return size / 1024
    
def fullwhite(png_name):
    im = Image.open(png_name)
    x,y = im.size
    try:
        # 使用白色来填充背景
        # (alpha band as paste mask).
        p = Image.new('RGBA', im.size, (255,255,255))
        p.paste(im, (0, 0, x, y), im)
        p.save(png_name)
    except:
        pass

def removebg():
    try:
        request_url = "https://aip.baidubce.com/rest/2.0/image-classify/v1/body_seg"
        # 二进制方式打开图片文件
        f = open(name, 'rb')
        img = base64.b64encode(f.read())
        params = {"image":img}
        access_token = get_access_token()
        request_url = request_url + "?access_token=" + access_token
        headers = {'content-type': 'application/x-www-form-urlencoded'}
        response = requests.post(request_url, data=params, headers=headers)
        if response:
            res = response.json()["foreground"]
            png_name=name.split('.')[0]+".png"
            with open(png_name,"wb") as f:
                data = base64.b64decode(res)
                f.write(data)
            fullwhite(png_name)
            png_jpg(png_name)
            os.remove(png_name)
            print(name+"\t处理成功!")
    except Exception as e:
        pass

if __name__ == '__main__':
    for i in paths:
        name = os.path.basename(i.name)
        if(name==os.path.basename(__file__)):
            continue
        size = get_size()
        ##照片压缩
        while size >=900:
            size = get_size()
            resize()   
        removebg()
        print(" ")

[Important] Precautions before use

1. This program will overwrite the original file. Before use Please back up the files to avoid data loss 2. Copy the program to the same directory as the photos to be processed, double-click the program to run

Guide you step by step to implement cutout and change the background color through the Python calling interface

Final rendering

Original image:

Guide you step by step to implement cutout and change the background color through the Python calling interface Rendering

Guide you step by step to implement cutout and change the background color through the Python calling interface

##Summary

The code is not difficult, but there are many small problems along the way. For example, the image size cannot exceed 4MB, and you have to compress the photo, path and other issues. In short, this function has been achieved. Very happy!

Okay, today’s sharing ends here ~


[Related recommendations:

Python3 video tutorial

]

The above is the detailed content of Guide you step by step to implement cutout and change the background color through the Python calling interface. For more information, please follow other related articles on the PHP Chinese website!

Statement:
This article is reproduced at:csdn.net. If there is any infringement, please contact admin@php.cn delete