Login

AI Engineering with Python and TensorFlow for Image Applications

This course introduces students to advanced image processing techniques using Python and TensorFlow. Students will learn how to develop state-of-the-art classifiers and segmentation models for various image applications. The course covers multi-class classification, binary classification, semantic segmentation, and object detection, along with a deep dive into the theory and intuition behind these techniques.
Fees: $1,900
 View Discount and Referral Programs 

Course Schedule

Course Start Date: Dec 1, 2024
Duration: 8 weeks
Location:  2500 Yale St Suite B24, Houston, TX 77008

Required Tools

A computer with Python 3.x and TensorFlow installed.
Access to a GPU for training deep learning models is recommended.

Certificate

Certificates will be awarded to the participants upon successful completion of the course requirements and passing the course exam. The certificate includes 18 PDH credits.

Course Objectives

  1. Understand the principles of deep learning for image applications using Python and TensorFlow.
  2. Develop state-of-the-art classifiers and segmentation models for various tasks.
  3. Explore the theory and intuition behind Convolutional Neural Networks (CNNs), semantic segmentation, and object detection.
  4. Learn how to prepare datasets for training classifiers and segmentation models.

Course Summary

Part 1: Multi-Class Classification with State-of-the-Art Architectures
  • Week 1: Building an Image Binary Classifier with ResNet
  • Week 2: Preparing a Dataset for Training a Classifier
  • Week 3: The Intuition and Theory of CNNs and Code Building Blocks
  • Week 4: State-of-the-Art Models for Image Classification
Part 2: Multi-Class Instance Segmentation with State-of-the-Art Architectures
  • Week 5: Building a Semantic Segmentation Model for Binary Masks
  • Week 6: The Intuition and Theory of Semantic Segmentation and Code Building Blocks
  • Week 7: Object Detection, theory, and intuition
  • Week 8: State-of-the-Art Models for Multi-Class Segmentation and Object Detection

Instructor

Dr. Nima Hamidi is a distinguished professional with a unique blend of expertise in music and computer science. He holds a Ph.D. in Music, showcasing their passion for the arts, and a Master's degree in Computer Science, highlighting their strong technical skills.
As the Computer Vision Lead at Bayer, Dr. Hamidi employs their extensive knowledge in data-driven techniques, machine learning, and artificial intelligence to develop innovative computer vision solutions. With years of experience as a Data Scientist, he has honed their skills in solving complex problems and delivering value to the organization.
In addition to their professional accomplishments, Dr. Hamidi has a passion for teaching and enjoys sharing their knowledge and experience with others. Their interdisciplinary background enables them to bring a fresh and innovative perspective to their work and teaching.

Week-by-Week Breakdown

Part 1: Multi-Class Classification with State-of-the-Art Architectures

Week 1: Building an Image Binary Classifier with ResNet
  • Introduction to TensorFlow
  • ResNet architecture
  • Implementing a binary classifier using ResNet
Week 2: Preparing a Dataset for Training a Classifier
  • Data acquisition and preprocessing
  • Data augmentation techniques
  • Training, validation, and test set preparation
Week 3: The Intuition and Theory of CNNs and Code Building Blocks
  • Convolutional layers and feature maps
  • Pooling layers and activation functions
  • Loss functions and optimization algorithms
Week 4: State-of-the-Art Models for Image Classification
  • Inception (GoogLeNet)
  • VGG
  • MobileNet
  • EfficientNet

 

Part 2: Multi-Class Instance Segmentation with State-of-the-Art Architectures

Week 5: Building a Semantic Segmentation Model for Binary Masks
  • Introduction to semantic segmentation
  • Implementing a binary mask segmentation model
  • Evaluation metrics for segmentation
Week 6: The Intuition and Theory of Semantic Segmentation and Code Building Blocks
  • Fully Convolutional Networks (FCNs)
  • U-Net and its variants
  • Loss functions and optimization algorithms for semantic segmentation
Week 7: Object Detection, theory, and intuition
  • Bounding box representation
  • Intersection over Union (IoU) metric
  • Single Shot MultiBox Detector (SSD)
  • You Only Look Once (YOLO)
Week 8: State-of-the-Art Models for Multi-Class Segmentation and Object Detection
  • Mask R-CNN
  • EfficientDet
  • DeepLabv3+
  • PANet