xinyooo/deepfake-detection

DeepFake Detection: Detect the video is fake or not using InceptionResNetV2.

deepfake_detection

Detect the video is fake or not using InceptionResNetV2.

The dataset was downloaded from kaggle deepfake detection challenge: https://www.kaggle.com/c/deepfake-detection-challenge/data

Experimental steps

Step1. run capture_img -> use the dlib face detecter to convert video to face img.

Step2. run deepfake_detection_train -> training the dataset from capture_img and output the model file.

Step3. run model_play -> load model then input video to detect the video is fake or not. [0]: fake, [1]: real

deepfake-detection's Issues

unable to run capture_img.ipynb

Hi I have problem with is can't run capture_img.ipynb the error show KeyboardInterrupt and also for the train_frame_folder what folder that need to use? need to use this format content/drive/MyDrive/train_sample_videos' ? also same with the metadata.json. I need your help for my final project. thank you.

i am getting FileNotFound and os not defined error, can u please resolve it?

environment

if possible, can u specify which python version,and the environment u used for this project?
this is for university research project only.

Issue while trying to train the model on data.

Hi,
After converting the videos into images and split them into fake and real folders then get our training data and compile the model, i wanted to train the model however an issue appears "None values not supported". I've checked if my input carried none values which wasn't the case, i made sure that i've run the model and it was compiled after getting the summary of the model as a result so please if you can check with me the problem and give me a solution in case you find it.

NB: In the word file you can find the flow of the script execution and the error attached.
Flow_of_script_running_and_errors.docx

about dataset

heloo, about the metadata in the training sample videos file,
the highlighted video name, should be the deepfake videos,
and the circled video name, should be the corresponding original one right?
am i right?

im confused because the the circled video name, does not exist in the sample video file

thank you in advance

Error: Empty Training Data

Hi, I am using my own dataset and the error seems that it does not read my training samples that I created from capture_img.ipynb . Is it because of the dataset directory or what? Btw, I am using google colab to run it.

Does this work?

I noticed that you had mentioned an over-fitting issue in your readme file earlier. I tried the model recently (yesterday) and I'm getting the same issue even after varying data-set size and epochs. I used the Kaggle deep-fake detection challenge data-set and followed the steps mentioned here. Could you tell me whether this model works on your datasets or about anything I might be doing wrong?

Unable to output img

Hi, I have cloned your code for capture_img.ipynb and I don't see any image file outputed by the program, can you help me, thank you so much.

dataset

do you have the training dataset?
because it is no longer available on the website stated

I am unable to run capture_img.py file. I am getting bad memory error.what should i do now..?