Diffusion Based Image Generation with Diffuser (Full Course) - Celebrity Face Generation Using Diffusion Models | By Dr Mohan Dash

#stablediffusion #aiart #python #generativeai #diffusion #pytorch I am Dr. Balyogi Mohan Dash, and I welcome you to Intelligent Machines! In this video, you will learn: Diffusion Model Training – Step-by-step training of a conditional diffusion model to generate celebrity faces from random noise using Hugging Face Datasets and Diffusers Library. Dataset Setup – Utilizes a 10,000-image celebrity face dataset (filtered to 1,000 samples), labeled by gender (male/female), preprocessed to 64×64 RGB tensors for faster training. Model Architecture – Implements a U-Net with class embeddings (2 classes), 17M parameters, trained using DPM Scheduler with DDIM-based generation for faster inference. Training Pipeline – End-to-end PyTorch workflow: noise sampling, forward diffusion, reverse denoising, 50K training steps, and visualization of generated samples at checkpoints. Results & Next Steps – Achieves realistic gender-conditioned faces; upcoming episode will cover image inpainting using diffusion models for targeted image editing. All code used in this video is available here: https://github.com/mohan696matlab/Diffusion_Gen_AI_Course Subscribe to the channel to follow this complete course and master diffusion models from theory to practical implementation. 🔗Links🔗 LinkedIn: https://www.linkedin.com/in/balyogi-mohan-dash/ GitHub: https://github.com/mohan696matlab Google Scholar: https://scholar.google.com/citations?user=jzcIElIAAAAJ&hl=en WhatsApp inquiries are currently closed. Please reach out via email for any questions: [email protected] 00:00 Introduction – Diffusion models and series overview 00:32 Goal – Generate celebrity faces using diffusion models 00:49 Dataset Setup – Hugging Face dataset with gender labels 01:41 Data Preprocessing – Resizing, normalization, and batching 03:39 Model Setup – U-Net architecture and scheduler configuration 05:56 Training Process – Noise prediction and sample generation 07:10 Results & Next Steps – Realistic faces and upcoming inpainting