This is the starter repository for the Weights & Biases community benchmark for adding realistic/natural color to black & white photos.
- Be sure to sign up for W&B.
- Clone this repository:
git clone https://github.com/wandb/colorizer-applied-dl.git - You can run the model and log the results to Weights & Biases by running:
$ pip install -U -r requirements.txt
$ wandb login
$ python color.py
Run python color.py to train a baseline Keras model that colorizes images of flowers.
Modify this file and the data pipeline (or write your own scripts and create different model architectures!) to get better results.
- Submit your results to the benchmark.
The goal is to add realistic/natural color to a given black and white photo.
We use a perceptual distance metric on the validation set to rank results (lower values are better).
This metric is generated by the color.py script. You are welcome to use other ML frameworks besides Keras, but please make sure to log the same perceptual distance metric to
W&B so we can rank results.
You can submit your best runs to our benchmark. More specifically, go the "Runs" table in the "Project workspace" tab of your project. Hover over the run's name, click on the three-dot menu icon that appears to the left of the name, and select "Submit to benchmark".
- Fancier architectures
- Different loss functions
- Data augmentation
- More training data?