Link to published research paper: https://ijeces.ferit.hr/index.php/ijeces/article/view/1719

Deepfake_Xception_Inception_resnet_v2_CNN

DeepFake-Detection-Project

We are trying to detect deepfake videos. Here we are using a dataset from Kaggle where the annotations are stored in a json file in the train sample videos folder.

Developed an ML model using YOLOv7 to identify warning signal lights in car dashboards, achieving 98% accuracy through bagging techniques and varied backbone architectures. Collaborated with cross-functional teams to integrate the model into a mobile app, improving real-time detection accuracy by 30%. Implemented testing suites and agile processes, reducing bugs by 40% and increasing user satisfaction by 25%.

Steps:

Reading the videos and taking images from it

Reading the label in the json file and store the image in a folder according to it's label

Converting the image to an array and splitting the data into train and test

Customizing the Xception, Inception_resnet_v2 and training the data on it

Testing