This GitHub repository showcases my bachelor thesis which is focused on exploring the application and comparison of various deep generative models for synthetic image augmentation in manufacturing domain.

Diffusion based generative model to synthesize 2d images from random noise using iterative denoising.

Implementation of the original denoising diffusion probabilistic model (DDPM) paper in PyTorch.

This repository contains my final submission for the COMP3547 Deep Learning module assignment at Durham University in the academic year 2022/2023. The project focuses on diffusion-based models and their application in synthesising new, unique images, which could plausibly come from a training data set. Final grade received was 71/100.

Simple Denoising Diffusion Probabilistic Model for 2d data points

Une série de notebooks qui expliquent en détail comment fonctionnent les modèles de diffusion

Python code for Colab for training the denoising diffusion model with Pytorch. Made with @PhilipTamb

[ICCV 2023] Official PyTorch implementation of the paper "DiffTAD: Temporal Action Detection with Proposal Denoising Diffusion"

Neural Architecture Search (NAS) on a UNet neural network (SR3) to optimize for minimal FLOPs and inference latency; while working in a denoising diffusion generative framework (DDPM)

Training and sampling from a denoising diffusion model on the MNIST dataset

[MICCAI2024 Early Accept] LighTDiff: Surgical Endoscopic Image Low-Light Enhancement with T-Diffusion

PyTorch implementation for DDPM & DDIM

An adventure through Inpainting hidden part of vehicles with Deep Learning!

Experiment with latent diffusion models.

"RecDiff: Diffusion Model for Social Recommendation"

Implementation of Denoising Diffusion Probabilistic Model from Scratch for Image Generation Task in PyTorch

Pytorch Implement of diffusion model

Implementation of a Denoising Diffusion Probabilistic Model with some mathematical background.

Custom implementation of the denoising task to generate dog faces 🐶 (experiment)

♨️ From scratch implementations of Denoising Diffusion Probabilistic Models for small toy examples to be run on local machines within minutes.