agusgun / fakeimagedetector Goto Github PK

untuk dataset .csv parameter yang diambil apa saja ?

Taking the project to an organized directory structure can help developers use this model and write relevant APIs. Kudos to the amazing project. Accuracy is not bad too

Hello,

What is the final of your program, because you don't share the results of your tests?

Thank you

hello agusgun, thank you for releasing the code, I try to download the datasets you used in the method from this link http://forensics.idealtest.org/ but there is no dataset at all...Could you please tell me how to get the datasets? Thank you very much~

I downloaded this dataset https://www.kaggle.com/sophatvathana/casia-dataset
but it has no any description, and some strange folder tree

├───CASIA1
│ ├───Au
│ ├───ela
│ └───Sp
├───CASIA2
│ ├───Au
│ └───Tp
└───__MACOSX
==├───CASIA1
==│ ├───Au
==│ └───Sp
==└───CASIA2
====├───Au
====└───Tp

which one should i use for train, which for test? which one is original pic which one is modified?
also i know the csv format

file_name,1 or 0 (fake or real image)
example for real image:
'datasets/train/real/Au_ani_00001.jpg',0

but i have no idea which folder should i use for source...

I appreciate for your great work, and I want repeat it by myself :- )

=========================================
so. i made this

I upload zip to google drive
unzip it to '/content/gdrive/My Drive/casia_dataset/
in google colab i generated csv by following code

is it right?

import os
path_orig = '/content/gdrive/My Drive/casia_dataset/CASIA2/Au/'
path_modif = '/content/gdrive/My Drive/casia_dataset/CASIA2/Tp/'

folder_orig = os.listdir()
folder_modif = os.listdir()

strings = []

for file in os.listdir(path_orig):
  try:
    if file.endswith('jpg'):
      if int(os.stat(path_orig + file).st_size) > 10000:
        line =  path_orig + file  + ',1\n'
        strings.append(line)
  except:
    print(path_orig+file)

for file in os.listdir(path_modif):
    try:
      if file.endswith('jpg'):
         if int(os.stat(path_modif + file).st_size) > 10000:
            line =  path_modif + file + ',0\n'
            strings.append(line)
    except:
      print(path_modif+file)

for line in strings:
      with open('/content/gdrive/My Drive/casia_dataset/dataset.csv', 'a') as f:
         f.write(line)

I used other dataset to run your program, but it's acc was much lower than yours. Why?

can you share the dataset ?

Hi @agusgun Thanks for sharing your work. Would you mind helping me with some of my queries?

After going through the work, I learned that the CNN learns to predict a categorical ELA image of the input image. Am I correct?
If so, since we have already know how to convert an input image to an ELA image using the function convert_to_ela_image(), i. e., we can calculate the 'ground truth' directly, may I know the idea of having a CNN model to do the conversion? Thanks

can u please tell what is in dataset.csv

Hello, I have a small doubt actually when I am trying to convert image into ELA i.e
"convert_to_ela_image('datasets/train/real/Au_ani_00001.jpg', 90)" I am getting an error that permission is denied. Can you please help where I am going wrong.

To be more specific about my error I am pasting my error below,

***** PermissionError: [Errno 13] Permission denied: 'C:\Users\saraiya.resaved.jpg' *****

Any help will be highly appreciated thanks in advance.

Hi!
Thank you for your works. Could you provide the pre-trained model? Thanks!

hello agusgun, thank you for releasing the code, I try to download the datasets you used in the method. But there is no link to the dataset. Please tell me how can I get that dataset.