Introduction to Artificial Intelligence for Image Recognition

Description:

In today’s AI-driven world, there is a growing expectation that enterprise software should be just as “smart” as the consumer apps we use every day. From facial recognition on smartphones to object detection in autonomous vehicles, image recognition is one of the most powerful and widely adopted applications of Artificial Intelligence.

This one-day training course provides a comprehensive introduction to the use of AI for image recognition. Participants will explore how neural networks—especially Convolutional Neural Networks (CNNs)—are designed to process visual data. Through practical labs and real-world examples, attendees will gain hands-on experience using TensorFlow and Keras to build, train, and evaluate image classification models.

Whether you're a developer, data scientist, or business professional looking to understand the foundations of AI for computer vision, this course offers the essential knowledge and skills to get started.

Duration: 1 Day

Course Code: BDT482

Learning Objectives:

By the end of this session, participants will be able to:

1. Understand the core principles of neural networks and convolutional neural networks (CNNs) used for image recognition tasks.
2. Gain practical experience using TensorFlow and Keras to build, train, and evaluate deep learning models for image data.
3. Explore key techniques in model optimization and tuning, including dropout layers, filters, batch size, and other hyperparameters.

• Familiarity with any programming language
• Recommended but not required:
  • Basic familiarity with the Python programming language

• Software Architects
• Developers

Course Outline:

• Module 1: AI Overview

  • A brief overview of AI
  • Machine Learning and Deep Learning, and their relationship to AI
  • Applying models for prediction

   

  Module 2: Python Essentials

  • A brief review of the essentials of Python for deep learning
  • Introduction to Jupyter Notebooks

   

  Module 3: Understanding Neural Networks

  • Neural network architecture
  • Learned weights
  • Activation functions
  • Backpropagation
  • Gradient Descent

   

  Module 4: Convolutional Neural Networks (CNN)

  • Kernels/Filters
  • Convolutional layer
  • Pooling layer
  • Fully-connected layer

   

  Module 5: AI with TensorFlow and Keras

  • Introducing TensorFlow/Keras
  • Sequential object
  • Flatten layer
  • Dense layers
  • Conv2D layers
  • Dropout layers
  • Lab: Setting up and Running TensorFlow
  • Lab: Build a Fully-Connected Neural Network with TensorFlow/Keras
    • Compile model
    • Optimizers
    • Loss functions
    • Metrics
  • Dataset handling
    • Training dataset
    • Validation dataset
    • Testing dataset

   

  Module 6: Building CNN Models

  • Lab: Crafting a CNN Model using TensorFlow/Keras
    • Hyperparameter tuning
    • Filters
    • Layers
    • Nodes
    • Batch size

   

  Module 7: Model Evaluation and Optimization

  • Lab: Evaluate the Learning Curve to Improve the Model
    • Learning curve
    • Epochs
    • Accuracy
    • Loss
    • Underfitting/Overfitting

   

  Training Content Provided: Yes (Digital Format)