DarkGeekMS/portrait-synthesis-from-speech

Issues

• Complete the graduation project book and final presentation

  #44 opened 3 years ago by DarkGeekMS
  0

• Complete the required project videos

  #45 opened 3 years ago by DarkGeekMS
  0

• Implement multiple head poses generation pipeline

  #41 opened 3 years ago by DarkGeekMS
  0

• Complete final speech recognition training

  #46 opened 3 years ago by DarkGeekMS
  0

• Re-train BERT multi-label text classifier for facial attributes

  #21 opened 3 years ago by DarkGeekMS
  0

• Decide speech recognition workflow

  #10 opened 3 years ago by DarkGeekMS
  0

• Re-structure the repository after completion

  #43 opened 3 years ago by DarkGeekMS
  0

• Explore alternative methods to BERT for text processing

  #40 opened 3 years ago by DarkGeekMS
  0

• Try StyleGAN2 fine-tuning for better feature separation

  #42 opened 3 years ago by DarkGeekMS
  0

• Setup normalizing flows framework for conditioned latent space manipulation

  #27 opened 3 years ago by DarkGeekMS
  0

• Expand the current set of feature directions through various method

  #23 opened 3 years ago by DarkGeekMS
  0

• Decide multiple head poses generation workflow

  #9 opened 4 years ago by DarkGeekMS
  1

• Perform final integration and sanity check of directions-based face generation

  #39 opened 4 years ago by DarkGeekMS
  0

• Adapt StyleGAN2 face generation workflow to face refinement module

  #37 opened 4 years ago by DarkGeekMS
  0

• Transfer StyleGAN2 to PyTorch in directions-based face generation

  #36 opened 4 years ago by DarkGeekMS
  0

• Explore possible improvements for end-to-end text-to-face generation network

  #17 opened 4 years ago by DarkGeekMS
  1

• Decide face refinement workflow

  #8 opened 4 years ago by DarkGeekMS
  1

• Text2Face dataset processing and expansion

  #12 opened 4 years ago by DarkGeekMS
  0

• Explore pretrained embedding to latent vector mapping

  #19 opened 4 years ago by DarkGeekMS
  1

• Re-calculate feature components of latent vector after each latent navigation

  #34 opened 4 years ago by DarkGeekMS
  1

• Generate face from text with centralized pose and neutral facial expression

  #28 opened 4 years ago by DarkGeekMS
  0

• Centralize skewed feature directions

  #35 opened 4 years ago by DarkGeekMS
  1

• Integrate generated feature directions to current text-to-face generation

  #30 opened 4 years ago by DarkGeekMS
  2

• Train a multi-label classification network for facial attributes

  #20 opened 4 years ago by DarkGeekMS
  0

• Apply orthogonalization to generated feature directions

  #31 opened 4 years ago by DarkGeekMS
  2

• Define a set of quantitative metrics to assess the system performance

  #26 opened 4 years ago by DarkGeekMS
  0

• Add debugging structure to directions-based text-to-face generation pipeline

  #32 opened 4 years ago by DarkGeekMS
  0

• Handle difference in magnitudes of generated feature directions

  #33 opened 4 years ago by DarkGeekMS
  0

• Define a complete set of facial features to work on for text-to-face generation

  #25 opened 4 years ago by DarkGeekMS
  0

• Disentangle generated feature directions

  #29 opened 4 years ago by DarkGeekMS
  1

• Add nonlinear weighting of differentiated logits to directions-based face generation

  #24 opened 4 years ago by DarkGeekMS
  1

• Complete directions-based staged face generation codebase

  #22 opened 4 years ago by DarkGeekMS
  0

• Implement text-to-face generation via attribute disentanglement

  #18 opened 4 years ago by DarkGeekMS
  1

• Load the complete end-to-end text-to-face generation network for training

  #13 opened 4 years ago by DarkGeekMS
  1

• Replace NLP module with more complicated approaches

  #7 opened 4 years ago by DarkGeekMS
  1

• Refactor the end-to-end network codebase to accept generic sentence embedding models

  #15 opened 4 years ago by DarkGeekMS
  0

• Write a comprehensive review about latent space navigation in StyleGAN2

  #14 opened 4 years ago by DarkGeekMS
  1

• Try open-source StyleGAN2 projectors

  #4 opened 4 years ago by DarkGeekMS
  1

• Complete a test script for the whole network

  #1 opened 4 years ago by DarkGeekMS
  0

• Use more robust loss for reconstructed images

  #6 opened 4 years ago by DarkGeekMS
  1

• Create a mapping network from StyleGAN2 latent space to FaceNet encoding

  #11 opened 4 years ago by DarkGeekMS
  1

• Improve training GPU utilization

  #2 opened 4 years ago by DarkGeekMS
  0

• Replace MSE loss for latent vectors with other losses

  #5 opened 4 years ago by DarkGeekMS
  1

• Add an option for loading pre-trained weights before training

  #3 opened 4 years ago by DarkGeekMS
  0