sns matplotlib output slicing the saved image

Question

I am trying to save sns matplotlib output as jpg file and reopen it with cv2.

but i am facing distinct data loss, would someone help me to resolve, i tried in several savefig options and documentations.

sample code

import pandas as pd
import numpy as np
import cv2
import seaborn as sns
import matplotlib.pyplot as plt
by_c = None

fig = plt.Figure(figsize=(5, 4), dpi=100)
g = sns.FacetGrid(pd.DataFrame(np.random.random(10)*150, columns=['col']), col=None,  row=None, height=3.5, aspect=1)
g.map_dataframe(sns.histplot, x='col')
plt.title('col'+' - '+str(by_c)+'-', fontsize=12)
plt.xlabel('Value')
plt.ylabel('Frequency')
plt.savefig('temp.png')
plt.show()

Out:

Saved picture example of 'temp.png'

out:

reopening image

im = cv2.imread('temp.png')
plt.imshow(im)

Out1: Image title and lables sliced bit, i am not sure how else i can save it. Would someone please help to resolve it

to rule out problems in cv2: can you post temp.png in the question as well? — LudvigH, Sep 01 '21 at 15:42
Actually, temp.png is also same @LudvigH, i tried to open it, opencv is simply opening what is saved :-| — Naga kiran, Sep 01 '21 at 15:52
ok. super. I think you can remove cv2 from the question altogether then... — LudvigH, Sep 01 '21 at 15:57
@LudvigH, Thanks for support , link really shown exact ans, seems i might not checked properly — Naga kiran, Sep 01 '21 at 16:03

Roxy · Answer 1 · 2021-09-01T16:04:14.587

1

To set the quality of the image use the dpi, and also specify the bbox_inches for a full layout. If not, it will consider the nearest view of the image

import pandas as pd
import numpy as np
import cv2
import seaborn as sns
import matplotlib.pyplot as plt
by_c = None

fig = plt.Figure(figsize=(5, 4), dpi=100)
g = sns.FacetGrid(pd.DataFrame(np.random.random(10)*150, columns=['col']), col=None,  row=None, height=3.5, aspect=1)
g.map_dataframe(sns.histplot, x='col')
plt.title('col'+' - '+str(by_c)+'-', fontsize=12)
plt.xlabel('Value')
plt.ylabel('Frequency')
plt.savefig('temp.png',dpi=300, bbox_inches = "tight")
#plt.savefig('temp.png')
plt.show()

im = cv2.imread('temp.png')
plt.imshow(im)

Resultant Image:

edited Sep 01 '21 at 16:04

answered Sep 01 '21 at 15:58

Roxy

1,015
7
20

this sure should fix the issue. but we should try to flag duplicates rather than answering them. also - please tidy code in answers so they are actually good code samples. (e.g. remove unnecessary cv2 dependency – LudvigH Sep 01 '21 at 16:01
Thanks for the update. Done as said. :) – Roxy Sep 01 '21 at 16:02
1

Thanks for the answer @Roxy, it solved the problem :) – Naga kiran Sep 01 '21 at 16:04
If it resolves, could you accept it as an answer maybe? – Roxy Sep 01 '21 at 19:26

sns matplotlib output slicing the saved image

1 Answers1