https://github.com/wiktorlazarski/edge-aware-face-autoencoder

👨‍🎨 Edge-aware Face Autoencoder.
https://github.com/wiktorlazarski/edge-aware-face-autoencoder

autoencoder celeba-dataset computer-vision deep-learning face generative-model hydra image-compression pytorch pytorch-lightning vae-pytorch wandb

Last synced: 2 months ago
JSON representation

👨‍🎨 Edge-aware Face Autoencoder.

• Host: GitHub
• URL: https://github.com/wiktorlazarski/edge-aware-face-autoencoder
• Owner: wiktorlazarski
• License: other
• Created: 2022-03-29T15:28:30.000Z (about 3 years ago)
• Default Branch: master
• Last Pushed: 2022-06-07T15:58:59.000Z (almost 3 years ago)
• Last Synced: 2025-01-19T06:19:59.127Z (4 months ago)
• Topics: autoencoder, celeba-dataset, computer-vision, deep-learning, face, generative-model, hydra, image-compression, pytorch, pytorch-lightning, vae-pytorch, wandb
• Language: Jupyter Notebook
• Homepage: https://colab.research.google.com/drive/1b3afUvKmuWQblkxIpGxUAOt0pZk3DBa3?usp=sharing
• Size: 93.1 MB
• Stars: 1
• Watchers: 3
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
______________________________________________________________________



# 👨‍🎨 Edge-aware Face Autoencoder



  👨‍🎓 Wiktor

  👩‍🎓 Aneta



______________________________________________________________________

[![ci-testing](https://github.com/wiktorlazarski/face-vae/actions/workflows/ci-testing.yml/badge.svg?branch=master&event=push)](https://github.com/wiktorlazarski/edge-aware-face-autoencoder/actions/workflows/ci-testing.yml)

[![Open In Collab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1b3afUvKmuWQblkxIpGxUAOt0pZk3DBa3?usp=sharing)

[![Code style: black](https://img.shields.io/badge/code%20style-black-000000.svg)](https://github.com/psf/black)



## 💎 Installation with `pip`

Installation is as simple as running:

```bash

pip install git+https://github.com/wiktorlazarski/edge-aware-face-autoencoder.git

```

## 🧐 Qualitative results

#### Original images

![alt text](https://github.com/wiktorlazarski/edge-aware-face-autoencoder/blob/master/doc/images/original_imgs.png)

---

#### Baseline model without edge-awareness trained on images with 256px resolution, VAE latent dimension equals to 512 and reconstruction loss weight equals to 100 000. Edges' weight was set to 1.

![alt text](https://github.com/wiktorlazarski/edge-aware-face-autoencoder/blob/master/doc/images/baseline-256px-512ld-100k.png)

---

#### Edge-aware model with edges' weight set to 3. All other parameters were the same as in the baseline model.

![alt text](https://github.com/wiktorlazarski/edge-aware-face-autoencoder/blob/master/doc/images/edgeaware-edge_weight3.png)

---

#### Edge-aware model with edges' weight set to 10. All other parameters were the same as in the baseline model.

![alt text](https://github.com/wiktorlazarski/edge-aware-face-autoencoder/blob/master/doc/images/edgeaware_edge_weight10.png)

---

## ⚙️ Setup for development with `pip`

```bash

# Clone repo

git clone https://github.com/wiktorlazarski/edge-aware-face-autoencoder.git

# Go to repo directory

cd edge-aware-face-autoencoder

# (Optional) Create virtual environment

python -m venv venv

source ./venv/bin/activate

# Install project in editable mode

pip install -e .[dev]

# (Optional but recommended) Install pre-commit hooks to preserve code format consistency

pre-commit install

```

## 🐍 Setup for development with `conda`

```bash

# Clone repo

git clone https://github.com/wiktorlazarski/edge-aware-face-autoencoder.git

# Go to repo directory

cd edge-aware-face-autoencoder

# Create and activate conda environment

conda env create -f ./conda_env.yml

conda activate face_autoencoder

# (Optional but recommended) Install pre-commit hooks to preserve code format consistency

pre-commit install

```



### 🤗 Enjoy the model !