Skip to content

KitPi/UNet

Repository files navigation

UNet

Creating a UNet Convolutional Neural Network for interactively colourising black and white photos.

Create Python virtual environment

python3 - m venv .venv

source .venv/bin/activate

pip install -r requirements.txt

Paper

Paper

Uni-Freiberg: Olaf Ronneberger: 18 May 2015

Debugger Cafe: Sovit Ranjan Rath: 3 April 2023

Modified to take greyscale photos as input, then output LUV colorised photos


Version 1

  • No BatchNorm
  • Image input and targets Normalized: $\mu$: 0.5, $\sigma$ 0.25
  • Epochs: 5
  • Mean Squared Error Loss
  • No dropout
  • Learning Rate: 5e-5

Training

Training loss (smoothed) / Time Training loss

Results

Results

Input Epoch 1 Epoch 2 Epoch 3 Epoch 4 Epoch 5 Ground Truth

Evaluation|300px


TODO


Version 2

Proposal

  • No normalisation
  • Improved Loss function: MSE + SSIM
  • BatchNorm
  • Skip Connection Convolution, then
  • Hintegration: Convolve hints, then integrate then convolve colour hints alongside down convolutions.
  • Dropout (not from school, from network)
  • Variable learning rate

About

A Convolutional Neural Network U-NET for colourising Black and White images.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages