Save the multiple images into PDF without chainging the format of Subplot

Question

I've a df like this as shown below. What I'm doing is I'm trying to loop through the df column(s) with paths & printing the image as sub plots one column with image paths at axis0 and other column paths parallely on axis1 as follows.

      identity       VGG-Face_cosine    img                 comment
0   ./clip_v4/3.png   1.110223e-16  .\clip_v3\0.png        .\clip_v3\0.png is matched with ./clip_v4/3.png
0   ./clip_v4/2.png   2.220446e-16  .\clip_v3\1.png        .\clip_v3\1.png is matched with ./clip_v4/2.png
1   ./clip_v4/4.png   2.220446e-16  .\clip_v3\1.png        .\clip_v3\1.png is matched with ./clip_v4/4.png
2   ./clip_v4/5.png   2.220446e-16  .\clip_v3\1.png        .\clip_v3\1.png is matched with ./clip_v4/5.png
0   ./clip_v4/2.png   2.220446e-16  .\clip_v3\2.png        .\clip_v3\2.png is matched with

I'm looping through these 2 columns identity and img columns & plotting as follows

import pandas as pd
import matplotlib.pyplot as plt
import matplotlib.image as mpimg
from matplotlib import rcParams


df = df.iloc[1:]
#merged_img = []


for index, row in df.iterrows():

    # figure size in inches optional
    rcParams['figure.figsize'] = 11 ,8

    # read images
    
    img_A = mpimg.imread(row['identity'])
    img_B = mpimg.imread(row['img'])

    # display images
    fig, ax = plt.subplots(1,2)
 
    
    ax[0].imshow(img_A)
    ax[1].imshow(img_B)

sample output I got.

###Console output

Upto now it's fine. My next idea is to save these images as it is with sublots on PDF. I don't want to change the structure the way it prints. Like I just want 2 images side by side in PDF too. I've went through many available solutions. But, I can't relate my part of code with the logic avaiable in documentation. Is there is any way to achieve my goal?. Any references would be helpful!!. Thanks in advance.

user_14909621 · Answer

Use PdfPages from matplotlib.backends.backend_pdf to save figures one by one on separate pages of the same pdf-file:

import pandas as pd
import matplotlib.pyplot as plt
import matplotlib.image as mpimg
from matplotlib import rcParams
from matplotlib.backends.backend_pdf import PdfPages

df = df.iloc[1:]
rcParams['figure.figsize'] = 11 ,8
pdf_file_name = 'my_images.pdf' 

with PdfPages(pdf_file_name) as pdf:

    for index, row in df.iterrows():
        img_A = mpimg.imread(row['identity'])
        img_B = mpimg.imread(row['img'])
        fig, ax = plt.subplots(1,2)
        ax[0].imshow(img_A)
        ax[1].imshow(img_B)

        # save the current figure at a new page in pdf_file_name
        pdf.savefig()

See also https://matplotlib.org/stable/api/backend_pdf_api.html

user_837451 · Answer

If you want all of the figures in one, you need to set up your axis in advance with the right number of subplots:

import pandas as pd
import matplotlib.pyplot as plt
import matplotlib.image as mpimg
from matplotlib import rcParams

df = pd.DataFrame(
    {
        "identity": ["0.png", "2.png", "4.png"],
        "img": ["1.png", "3.png", "5.png"],
    }
)

fig, ax = plt.subplots(len(df), 2, figsize=(11, 8*len(df)))

for [index, row], axrow in zip(df.iterrows(), ax):

    # read images
    img_A = mpimg.imread(row["identity"])
    img_B = mpimg.imread(row["img"])

    # display images

    axrow[0].imshow(img_A)
    axrow[1].imshow(img_B)

plt.savefig("all_images.png")
plt.close()

This is what the output looks like (input is a handful of stackoverflow user pics):

Save the multiple images into PDF without chainging the format of Subplot

2 Answers