StarNet Software

Running options

USAGE:

  deepsnr  [-ehq] [--version] [-m <int>] [-o <string>] [-s <int>]
           [-w <string>] -i <string>

Options:
  -i <string>, --input <string>
      Required input image filename. Recommended: TIFF/TIF or PNG.

  -o <string>, --output <string>
      Output image filename. TIFF/PNG default to 16-bit unless --eight is used.

  -w <string>, --weights <string>
      Override path to the model checkpoint.

  -s <int>, --stride <int>
      Stride value. Must be even and between 2 and 512.

  -m <int>, --model <int>
      Model version to use, 1 or 2. Default is 2. Model 1 requires RGB input.

  -q, --quiet
      Do quiet run with no terminal output.

  -e, --eight
      Write TIFF and PNG output as 8-bit instead of the default 16-bit.

  --, --ignore_rest
      Ignores the rest of the labeled arguments following this flag.

  --version
      Displays version information and exits.

  -h, --help
      Displays usage information and exits.

DeepSNR CLI, ONNX Runtime backend.

Inputs:
  Recommended: TIFF/TIF or PNG, 8-bit or 16-bit integer,
               grayscale or RGB.
  Might work: JPEG/JPG and BMP through OpenCV,
              but these formats were not tested.
  Not accepted: floating-point images, alpha channels,
                unsupported channel/depth layouts.

Models:
  Model 1: RGB/color images only.
  Model 2: RGB/color and grayscale/monochrome images.

  Both models work on images from monochrome CCD cameras.
  Drizzle-integrated images from one-shot color cameras might work as well.
  DeepSNR works best on uncorrelated high-frequency noise;
    correlated noise, such as walking noise, will yield poor results.

Outputs:
  TIFF and PNG outputs are saved as 16-bit or 8-bit with --eight.
  TIFF outputs are always saved with LZW compression.

Package layout:
  Portable archive:
    To run from the extracted archive, keep these files together:
    deepsnr/
      deepsnr or deepsnr.exe
      DeepSNR_weights_v1.onnx
      DeepSNR_weights_v2.onnx
      lib/ or runtime DLLs
      README.txt
      LICENSE.txt

  Installed package:
    Native installers place the executable in a system bin directory and keep
    model/runtime files in the matching DeepSNR support directory. See CLI
    installer locations for OS-specific paths.

DeepSNR v1.2.1

Example commands

Denoise with the default current model

deepsnr --input input.tif --output denoised.tif

Choose model 2 explicitly

deepsnr --input input.tif --output denoised.tif --model 2

Run the older RGB model line

deepsnr --input rgb-input.tif --output denoised.tif --model 1

Use a specific weights file

deepsnr --input input.tif --output denoised.tif --weights DeepSNR_weights_v2.onnx

Use quiet mode for scripts

deepsnr --input input.tif --output denoised.tif --quiet

Inputs and outputs

DeepSNR is intended for high-quality calibrated and integrated data where the remaining noise is mostly random and not strongly patterned. Correlated artifacts such as walking noise behave more like image structure and should be addressed before denoising.

File formats

TIFF/TIF and PNG are the recommended tested input formats. JPEG/JPG and BMP might work through OpenCV, but they were not tested for this release. JPEG is lossy and is not recommended for scientific or archival data.

Bit depth

8-bit and 16-bit integer inputs are supported. Unsupported input depths, including 32-bit floating-point images, are rejected; convert them to 16-bit integer before running the tool.

Color and grayscale

Model 1 accepts RGB/color images only. Model 2 accepts RGB/color and true grayscale or monochrome images. Images with alpha channels or other channel counts are rejected.

Image size

Images must be at least 512 × 512 pixels.

Default output depth

TIFF and PNG outputs are written as 16-bit integer images by default.

--eight

Use this only when you intentionally want 8-bit TIFF or PNG output.

Compression

TIFF outputs are always saved with LZW compression. PNG output uses the default OpenCV encoding. Other output formats are written as 8-bit images.

DeepSNR-specific options

--model, -m

Chooses the built-in model line when --weights is not provided. Model 2 is the default current model line and supports RGB/color and grayscale inputs. Model 1 is the older RGB-only model line.

Automatic weights lookup

When --weights is omitted, DeepSNR chooses DeepSNR_weights_v1.onnx or DeepSNR_weights_v2.onnx next to the executable based on --model.

--stride, -s

Tile stride. It must be even and between 2 and 512. DeepSNR defaults to 480.

Short option clustering

Short boolean switches can be combined. For example, -eq is equivalent to -e -q. Options that take values, such as -i, -o, -m, -w, and -s, must be provided separately.