Running support on patches of video

[bantin]

Hi Support folks --

I'd like to run SUPPORT on small patches of a voltage imaging video (say 10 pixels x 10 pixels x Time). Is it possible to do this? It seems like with the pretrained models they need a larger image patch. If I train my own model, can I control the size of the filters?

[SteveJayH]

Hi,

We apologize for the late reply.

Since the pretrained models are trained on large images, they may not perform well on small videos. So it is better to train the model on your specific data.

Due to the small spatial dimension, we need to modify the model. Specifically, reduce the depth to 3 and set the unet_channels as [16, 32, 64], as follows:

python -m src.train --exp_name YOUR_EXPERIMENT_NAME --noisy_data YOUR_PATH --depth 3 -- unet_channels 16 32 64 --bs_size 1 1 --patch_size 61 10 10 --patch_interval 1 1 1 --input_frames 61 -- checkpoint_interval 100

We checked that this setting works without errors for 10 x 10 pixels data, but we are not certain about the denoising performance on your data.

Please let us know if you have any further questions or issues.