vitalwarley/estagio-pdi

Recursos para as aulas do estágio docência na disciplina de Processamento de Imagens

Digital Image Processing: OCR with MLOps Using DVC and Guild AI

This project is part of a two-day practical workshop on Digital Image Processing, focusing on OCR (Optical Character Recognition) with an emphasis on MLOps practices using DVC (Data Version Control) and Guild AI. The workshop covers setting up an OCR pipeline, versioning datasets, running and tracking experiments, hyperparameter tuning, and analyzing results.

Table of Contents

1. Prerequisites
2. Day 1 Practice
   • Setting up the Environment
   • Project Structure and DVC Setup
   • Implementing Traditional OCR Pipeline
   • Running and Tracking Experiments with DVC
   • Homework
3. Day 2 Practice
   • Advanced DVC Usage
   • Defining Hyperparameter Spaces in DVC
   • Running a Hyperparameter Tuning Experiment
   • Analyzing and Visualizing Results
4. References

Prerequisites

Before starting, ensure you have the following installed:

• Python 3.11
• pip
• Git

Install required Python packages:

pip install -r requirements

Verify installations:

dvc doctor

References

• DVC Documentation
• OpenCV Documentation
• scikit-learn documentation