如何将图像文件从URL转换为scikit-learn中可用于执行多元线性回归的格式

Question

I am trying to do a simple multivatiate Linear Regressions where one of the features columns are images. I currently have the URL of each image. How do I convert this URL column into a format that scikit learn will understand(a numerical version of a picture)?

import pandas as pd
import numpy as np
from sklearn import linear_model

df = pd.read_csv('dtaset.csv')
df.head()

index followers  likes      urls
  0   14928252   132771   https://url1
  1   14928252   57671    https://url2
  2   14928252   161580   https://url3
  3   14928252   60969    https://url4
  4   14928252   375567   https://url5

The Feature columns are 'total_followers' and 'urls' and the target column is "likes"

Anyone? Thanks!

Answer 1

Your question has an easy part and a hard part. The easy one is how to read an image from URL. For this you will need OpenCV. If you don't have it ( import cv2 fails), install it, eg using pip install opencv-python . Than you can use this function:

import cv2
import requests
def readImage(url, color = False):
    with requests.get(url) as r:
        return cv2.imdecode(np.frombuffer(r.content, np.uint8), cv2.IMREAD_COLOR if color else cv2.IMREAD_GRAYSCALE)

And the hard question is what you will do with the images? An image is (usually) a large array of numbers, most probably, you will need to extract some characteristics and features from it in order to use them in a linear regression. But it depends on the nature of your task.