​
Model Training Overview

Welcome to the Model Training section of Ocular Foundry! This guide provides a high-level overview of what model training is, why it’s beneficial, and what you can achieve with it.

Model training lets you create a custom AI model specialized in detecting objects from your own images, leading to significantly higher accuracy for your specific use case.

​
What is Model Training?

Model training is the process of teaching an AI model to recognize specific patterns in your data. Instead of using a generic, one-size-fits-all model, you can train a custom, fine-tuned model using data curated and annotated with Ocular Founmdry.

​
The Journey: From Data to a Custom Model

The relationship between your data and your trained models follows this structure, ensuring that your experiments are organized and reproducible.

Ocular
  ├── Project
  │     ├── Dataset
  │     │     ├── Version
  │     │     │     ├── Model
  │     │     │     └── ...
  │     │     └── ...
  │     └── ...
  └── ...
  1. Project: The main container for your work.
  2. Dataset: A collection of your images within a project.
  3. Version: A snapshot of your dataset with specific annotations, from which you can train one or more models.

This hierarchical structure aligns with how you organize your data in the Ocular Foundry platform, making it intuitive to manage your ML workflows.

​
Core Benefits

  • Higher Accuracy: A custom-trained model will always outperform a generic one on your specific data.
  • Specialized Knowledge: The model learns the unique visual characteristics of the objects you care about.
  • Flexibility: Experiment with different configurations to find the perfect balance of speed and accuracy.

Ready to start training? Head over to the Model Training guide to begin creating your first model.