How to convert image into dataframe

#1

how can i convert image into a panda dataframe?

#2

What kind of DataFrame do you want to make from the image (i.e. the shape)?
Do you want the df of the shape: (width, height) or a linear one with the RGB values as columns?

#3

I want a linear one with RGB values as col

#4

This snippet should do the job:

colourImg = Image.open("test.png")
colourArray = np.array(colourImg.getdata())

df = pd.DataFrame(colourArray, columns=["red","green","blue"])
#5

it says np does not have any attribute getdata()

#6

from PIL import Image
Use this Image class for the Image.open() method.

#7

I already imported and modified the img…

#8

If the image is already in a numpy array, then you can also do this:

pd.DataFrame(img.reshape(-1, 3), columns=["red","green","blue"])

Ensure channels are the last dimension.

#9

all the values of the dataframe are zero
I don’t have any clue why
pls help me

#10

Hey @stutijain578
I need to look at the code to further investigate the issue. Request you to share it through Google Drive or the Coding Blocks IDE.

#11

https://drive.google.com/file/d/16KoS98R-uqx3Ywx4r6slCUf6oIi6Ap2Z/view?usp=sharing

#12

Hey Stuti.
The file is not visible. Please make it public.

#13

I have made it public

#14

Hi Stuti, you are mistaken. All rows are not zero. To confirm this, you can use any of these functions:

df.sum()
df.max()
df.std()
df[30000:30500]
#15

ok
thanks
btw my collage is bit different from that of sample even though I followed the instructions for the mentioned dimensions…

#16

Have you tried it submitting it? What is the score you received? Also try verifying the dimensions and see if they match.

#17

it is not getting uploaded, says can’t upload

#18

If it’s a purely upload issue, request you to drop a mail at [email protected]
They’ll help you.

#19

accuracy is only 19%, please mention the dimensions of center image

#20

Hi @stutijain578

The dimensions of the center image are (100, 100) without the borders. Try to make a submission after this correction.