Contents x
      Pre-registered docker images
      • 01 Mar 2023
      • 1 Minute to read
      • Print
      • Dark
        Light
      • PDF
      Contents

      Pre-registered docker images

      • Updated on 01 Mar 2023
      • 1 Minute to read
      • Print
      • Dark
        Light
      • PDF
      Article summary

      This table lists the pre-registered docker images available for use in pipelines.

      Name
      		Type
      		Details
      AkridataThumbnailThumbnail generator(Recommended) Produces a thumbnail with up to a resolution of 192x108 in JPEG format and aspect ratio preserved.
      AkridataImagePreprocessorImage Preprocessor(Recommended for most use cases) This is a pass-through pre-processor.
      AkridataMedicalImagePreprocessorImage PreprocessorPre-processor specifically suited for medical field data.
      AkridataFeaturizerFeaturizer(Full Image)(Recommended) This featurizer is suitable for use cases where a single object dominates an image, and there is no use for a patch search operation.
      AkridataPatchFeaturizerV2Featurizer(Patch)(Recommended) This featurizer produces features for each cell on a 7x7 grid over the image. This featurizer is suitable for use cases requiring patch search where objects of interest exist within a variety of surroundings
      AkridataSplitPatchFeaturizerFeaturizer(Patch)This featurizer splits the image into a 5x5 grid and produces features on each grid cell separately. The featurizer is about 10X slower than the recommended AkridataPatchFeaturizerV2 but produces features that are fully localized to each grid cell.
      AkridataSplitPatchFeaturizerSmallCellsFeaturizer(Patch)This featurizer splits the image into a 10x10 grid and produces features on each grid cell separately. The featurizer is about 4X slower than AkridataSplitPatchFeaturizer and is recommended only if there is a use case for running a patch search with very small objects being searched for.
      AkridataExternalFeaturizerFeaturizer(External)This featurizer must be used in pipelines where the features are provided through an external CSV file. 
      AkridataPatchFeaturizerFeaturizer(Patch)(Deprecated) This featurizer produces features for each cell on a 7x7 grid over the image and has been superseded by AkridataPatchFeaturizerV2.
      Was this article helpful?
      What's Next