Skip to main content

MLUtility Workflow Overview

The MLUtility module on the Gesund.ai platform provides a powerful set of tools for machine learning-driven dataset exploration and image analysis. From computing similarity between datasets to classifying individual pixels, MLUtility simplifies and accelerates key tasks in your data pipeline.

MLUtility Workflow

What Can You Do With MLUtility?​

MLUtility supports the following high-impact workflows:

  • Discover similarities across datasets
  • Run advanced dataset-level analyses
  • Annotate and enhance images
  • Train and use pixel classification models

Each step is accessible via UI or API endpoints, making integration flexible for both technical and non-technical users.

Step-by-Step Workflow​

1. Dataset Similarity​

  • Compare datasets based on content and image features
  • Initiate similarity calculations via API
  • Outputs include:
    • Similarity matrix
    • Matched dataset and image IDs
  • Helps in:
    • Curating training sets
    • Identifying duplicates or related studies

2. Dataset Analysis​

  • Trigger a dataset analysis run for any dataset
  • The engine generates:
    • Distribution plots
    • Statistical summaries
    • Metadata insights
  • Results are stored for re-use in downstream analysis
  • Benefits:
    • Understand dataset composition
    • Improve data curation decisions

3. Image Annotation​

  • Use image enhancement and annotation tools to clean and prepare images
  • Main feature:
    • Background removal using ML-powered filters
  • Use cases:
    • Preprocessing before segmentation
    • Normalizing diverse image sources

4. Pixel Classification​

  • Train models to classify each pixel in an image
  • Workflow:
    • Upload dataset and define class mapping
    • Train a pixel classifier
    • Apply model to new images for inference
  • Ideal for:
    • Object detection
    • Region-based analysis
    • Medical image segmentation

Why Use MLUtility?​

  • Save time on manual data processing
  • Improve the consistency and quality of image annotations
  • Leverage built-in ML tools without needing separate infrastructure
  • Increase accuracy in downstream models by cleaning and understanding your data first

MLUtility transforms complex image and dataset operations into a streamlined, user-friendly workflow on the Gesund.ai platform.