Comprehensive Filters

The comprehensive filters provide the option to process image data with deep learning models trained for tasks such as denoising, segmentation, and image enhancement. You should note that a selection of trained and untrained deep learning models are available for download in the Infinite Toolbox (see Infinite Toolbox). You will need to download or create at least one model before you can access the comprehensive filters.

The following examples provide the results of processing image data with deep learning models.

Original data (below left) and data denoised (below right)

Original data (below left) and data processed with a membrane detection model (below right)

Original data (below left) and data after processing with a segmentation model (below right)

You should note that downloaded models, as well as models that you created and trained for processing image data, will be available in the Options box of the Image Processing panel, circled below, whenever you choose either Deep Learning or Deep Learning Denoiser as a comprehensive filter.

Options for Deep Learning filters

Deep Learning denoising requires two inputs — the original data that you need to process and a variation of the original data that was processed with a comparable smoothing filter, such as a Gaussian or mean filter.

Deep Learning Denoiser filter inputs

You should note that the second input must be the same shape as the first.