Data_Augmentation_gan

Overview

This project focuses on frame interpolation using a convolutional neural network (CNN) for video processing. The goal is to predict intermediate frames between two consecutive frames in a video sequence, enhancing the smoothness of motion.

Project Structure

• Frame_interpolation.py: Contains the definition of the frame interpolation model (Data_Aug_Model).
• dataset.py: Defines the custom dataset class (Custom_Dataset) for loading and preprocessing frames.
• train.py: Implements the training script for the frame interpolation model.
• generate.py: Demonstrates how to use the trained model to generate interpolated frames.
• Pre_trained_Model/: Directory to store pre-trained model weights.
• Healthy/: Directory containing input frames for training and testing.
• results/: Directory to store the output frames generated by the model.

Requirements

• Python 3.x
• PyTorch
• torchvision
• tqdm
• Pillow

Install dependencies using:

pip install -r requirements.txt

Training

To train the frame interpolation model, run:

python train.py

Training

To generate interpolated frames using the trained model, run:

python generate.py