Overview

Ready to apply your AI skills in a real-world scenario you can showcase in your portfolio? During this project, you’ll work with the deep learning skills you’ve acquired throughout the Professional Certificate, and we recommend that you have completed all the previous courses before starting this one. For this project, you’ll build and compare deep learning models using Keras and PyTorch, and work through a full development pipeline from data loading and augmentation to model training, evaluation, and deployment. You’ll apply convolutional neural networks (CNNs) and vision transformers to domain-specific challenges. Then, finally, you’ll assess performance using metrics like accuracy, precision, and inference speed. By the end of the project, you’ll be able to demonstrate your skills in building and comparing models using Keras and PyTorch. Plus, you’ll be able to showcase that you can implement CNNs and vision transformers and evaluate your model’s performance. If you’re ready to complete a portfolio-worthy capstone project, enroll today!

Syllabus

Data Handling

This module lays the foundation for your capstone project by introducing the real-world case study you will work on. It also highlights essential prerequisites, including key concepts and tools required for deep learning development. You’ll explore data handling and augmentation and prepare your local development environment. Through hands-on labs using geospatial image data, you’ll build a custom geographical territory data loader system and gain experience with memory-based and generator-based data loading. The module also introduces Keras and PyTorch workflows, setting the stage for advanced model development in the later modules.

Convolutional Neural Network Model Development

In this module, you will dive into the practical implementation of convolutional neural networks for image classification tasks. Focusing on an agricultural land classification use case, you will build and train CNN models using Keras and PyTorch. Through hands-on labs, you’ll gain experience in constructing and optimizing models in each framework. The module concludes with a comparative analysis of the two approaches, helping you understand the trade-offs in model performance, training efficiency, and deployment considerations. By the end of this module, you’ll be equipped to choose the right framework for a given problem and justify your design decisions using real evaluation metrics.

CNN - Vision Transformer Integration

In this module, you will explore vision transformers, a cutting-edge deep learning architecture originally developed for natural language processing and now transforming the field of computer vision. You will learn how to apply transfer learning to vision transformers for real-world image classification tasks. Using PyTorch and Keras frameworks, you’ll implement, fine-tune, and compare vision transformer models in practical scenarios. You’ll perform comparative evaluation between the vision transformer performance for Keras-based and PyTorch-based models.

Final Project Submission and Course Wrap-Up

In this module, you will make a final submission of all the labs you’ve completed throughout the course for evaluation. Your submission will be AI graded. Finally, we will wrap up the course by highlighting key takeaways and outlining next steps.