Using plots/images directly within Streamlit app without saving them

[Robert Dzudzar]

Hi there,

Excellent Streamlit component, thanks for making it available!

I wanted to ask if there is a way of parsing directly figures from plot onto it, instead of using saved image files?

Example usage:
Having fig1, fig2 created in matplotlib within Streamlit app and then using fig1 and fig2 directly in streamlit-image-comparison?

Cheers,
Robert

[Robert Dzudzar]

Ok, so I found solution.
I've saved plot into BytesIO, and then using Image from PIL parsed it into the image_comparison.

Example:

import matplotlib.pyplot as plt
from PIL import Image
from io import BytesIO

x = [1,2,3]
y = [5,6,7]
z = [2,2,2]

plt.plot(x, y)
fig1 = BytesIO()
plt.savefig(fig1)

plt.plot(x, z)
fig2 = BytesIO()
plt.savefig(fig2)

image_comparison(
    img1=Image.open(fig1).convert('RGB'), #need to convert to RGB
    img2=Image.open(fig2).convert('RGB'),
    label1="Figure 1",
    label2="Figure 2"
)

[fatih c. akyon]

@rdzudzar awesome solution! How can we automate it inside image_comparison function?

Is it possible to come up with something like:

if isinstance(fig1, matplotlib.figure):
    buf = io.BytesIO()
    fig1.savefig(buf)
    buf.seek(0)
    img1_pillow = PIL.Image.open(buf)

[Robert Dzudzar]

Yes, possibly, adding some checks of what is parsed. I could check code as true with this modifications:

if isinstance(fig1, io.IOBase):
    buf = BytesIO()
    plt.savefig(buf)
    buf.seek(0)
    img1_pillow = Image.open(buf)

Since figure is within BytesIO, so its <class '_io.BytesIO'>.
fig1.savefig(buf) didn't work, so I replaced it with plt.savefig(buf).