Skip to main content

Optimizing Deep Learning Model Performance for Computer Vision


Department of Information and Communications Technology
Enrollment in this course is by invitation only

Course Overview

AI engineers and researchers in the academe and industry have discovered so many applications of Deep Learning in various problem domains such as natural language processing and computer vision. Specifically, in computer vision, Deep Learning has been used in object localization, detection, semantic segmentation, and pose estimation. Deep Learning-based approaches have been proven to surpass the performance of classical methods for computer vision on several occasions. However, a Deep Learning model’s performance is highly dependent on several factors, the most prominent one being the quantity and quality of its input dataset. In this course, we will look into improving the performance of Deep Learning models through data augmentation methods and transfer learning. This course will equip you in developing more performant Deep Learning models in your own AI engineering projects.

What You Will Learn

At the end of this course, you will be able to:

  • determine how to fine-tune their deep learning models using data augmentation methods
  • recognize how to fine-tune their deep learning models using pre-trained models

Course Content

Week 1: Data Augmentation for Training Neural Networks

7 Videos | 1 Activity

7 Videos

  • Welcome to the course!
  • Neural Network Performance and Data Augmentation
  • Data Augmentation in Computer Vision
  • Visualizing Data Augmentation
  • Dataset for Classification Using Deep Learning
  • Neural Network Performance with Data Augmentation
  • Summary

1 Activity

  • Exit Assessment

Week 2:Transfer Learning: Extracting Features Using Neural Networks

6 Videos | 1 Activity

6 Videos

  • What is Transfer Learning?
  • Feature Extraction Using Pre-Trained Neural Networks for Transfer Learning
  • Extracting Features from Image Datasets
  • Using Extracted Features to Train a Classifier Model
  • Summary

1 Activity

  • Exit Assessment

Week 3: Transfer Learning: Fine-Tuning Pre-trained Neural Network Models

5 Videos | 1 Activity

5 Videos

  • Pre-trained Neural Network Models
  • Identifying Layer Indices
  • Creating a New Network Top Layer
  • Training a Network with Replaced Top Layer
  • Summary

1 Activity

  • Exit Assessment

Week 4: Improving Neural Network Accuracy

9 Videos | 1 Activity

9 Videos

  • What are Ensemble Methods?
  • Why Do Ensemble Methods Typically Work?
  • Creating an Ensemble of Neural Network Models
  • Evaluating the Performance of an Ensemble of Models
  • Adagrad
  • Adadelta
  • RMSprop
  • Adam
  • Key Takeaways

1 Activity

  • Exit Assessment
  1. Course Number

    DICT-ICT004
  2. Classes Start

    TBA
  3. Estimated Effort

    2 hrs./week (8 hours)
  4. Price

    Free