Denoising

The goal is to remove noise from an image. Our library includes Noise2Void [KBJ19] using any of the U-Net versions provided. The main advantage of Noise2Void is neither relying on noise image pairs nor clean target images since frequently clean images are simply unavailable.

• Input:

  • Noisy image.

• Output:

  • Image without noise.

In the figure below an example of this workflow’s input is depicted:

Input image. Obtained from Noise2Void project.

Data preparation

To ensure the proper operation of the library the data directory tree should be something like this:

dataset/
├── train
│   └── x
│       ├── training-0001.tif
│       ├── training-0002.tif
│       ├── . . .
│       ├── training-9999.tif
└── test
    └── x
        ├── testing-0001.tif
        ├── testing-0002.tif
        ├── . . .
        ├── testing-9999.tif

Configuration file

Find in templates/denoising folder of BiaPy a few YAML configuration templates for this workflow.

Special workflow configuration

Please refer to Noise2Void to understand the method functionality. These variables can be set:

• PROBLEM.DENOISING.N2V_PERC_PIX controls the percentage of pixels per input patch to be manipulated. This is the n2v_perc_pix in their code.

• PROBLEM.DENOISING.N2V_MANIPULATOR controls how the pixels will be replaced. This is the n2v_manipulator in their code.

• PROBLEM.DENOISING.N2V_NEIGHBORHOOD_RADIUS controls the radius of the neighborhood. This is the n2v_neighborhood_radius in their code.

• PROBLEM.DENOISING.N2V_STRUCTMASK wheter to use Struct Noise2Void.

Run

Command lineDockerGoogle Colab

Open a terminal as described in Installation. For instance, using 2d_denoising.yaml template file, the code can be run as follows:

# Configuration file
job_cfg_file=/home/user/2d_denoising.yaml
# Where the experiment output directory should be created
result_dir=/home/user/exp_results
# Just a name for the job
job_name=2d_denoising
# Number that should be increased when one need to run the same job multiple times (reproducibility)
job_counter=1
# Number of the GPU to run the job in (according to 'nvidia-smi' command)
gpu_number=0

# Move where BiaPy installation resides
cd BiaPy

# Load the environment
conda activate BiaPy_env
source $CONDA_PREFIX/etc/conda/activate.d/env_vars.sh

python -u main.py \
    --config $job_cfg_file \
    --result_dir $result_dir  \
    --name $job_name    \
    --run_id $job_counter  \
    --gpu $gpu_number

For multi-GPU training you can call BiaPy as follows:

gpu_number="0, 1, 2"
python -u -m torch.distributed.run \
    --nproc_per_node=3 \
    main.py \
    --config $job_cfg_file \
    --result_dir $result_dir  \
    --name $job_name    \
    --run_id $job_counter  \
    --gpu $gpu_number

nproc_per_node need to be equal to the number of GPUs you are using (e.g. gpu_number length).

Results

The results are placed in results folder under --result_dir directory with the --name given. An example of this workflow is depicted below:

Example of denoising model prediction.

Following the example, you should see that the directory /home/user/exp_results/my_2d_denoising has been created. If the same experiment is run 5 times, varying --run_id argument only, you should find the following directory tree:

my_2d_denoising/
├── config_files/
│   └── my_2d_denoising.yaml
├── checkpoints
|   ├── my_2d_denoising_1-checkpoint-best.pth
|   ├── normalization_mean_value.npy
│   └── normalization_std_value.npy
└── results
    ├── my_2d_denoising
    ├── . . .
    └── my_2d_denoising
        ├── cell_counter.csv
        ├── aug
        │   └── .tif files
        ├── charts
        │   ├── my_2d_denoising_1_n2v_mse.png
        │   ├── my_2d_denoising_1_loss.png
        │   └── model_plot_my_2d_denoising_1.png
        ├── per_image
        │   └── .tif files
        ├── train_logs
        └── tensorboard

• config_files: directory where the .yaml filed used in the experiment is stored.

  • my_2d_denoising.yaml: YAML configuration file used (it will be overwrited every time the code is run).

• checkpoints: directory where model’s weights are stored.

  • my_2d_denoising_1-checkpoint-best.pth: checkpoint file (best in validation) where the model’s weights are stored among other information.

  • normalization_mean_value.npy: normalization mean value (only created if DATA.NORMALIZATION.TYPE is custom). Is saved to not calculate it everytime and to use it in inference.

  • normalization_std_value.npy: normalization std value (only created if DATA.NORMALIZATION.TYPE is custom). Is saved to not calculate it everytime and to use it in inference.

• results: directory where all the generated checks and results will be stored. There, one folder per each run are going to be placed.

  • my_2d_denoising_1: run 1 experiment folder.

    • cell_counter.csv: file with a counter of detected objects for each test sample.

    • aug: image augmentation samples.

    • charts:

      • my_2d_denoising_1_*.png: Plot of each metric used during training.

      • my_2d_denoising_1_loss.png: Loss over epochs plot (when training is done).

      • model_plot_my_2d_denoising_1.png: plot of the model.

    • per_image:

      • .tif files: reconstructed images from patches.

• train_logs: each row represents a summary of each epoch stats. Only avaialable if training was done.

• tensorboard: Tensorboard logs.

Note

Here, for visualization purposes, only my_2d_denoising_1 has been described but my_2d_denoising_2, my_2d_denoising_3, my_2d_denoising_4 and my_2d_denoising_5 will follow the same structure.