Birding with AI Concepts and Projects for Ornithology

About this book

Birding with AI introduces readers to the increasingly ubiquitous realm of artificial intelligence and its applications in ornithology and wildlife biology. As well as showcasing the potential utility of deep learning in ornithology, the book demonstrates how to understand, design, implement and evaluate AI models for ornithology and related fields.

Readers will learn:
- The background of AI, specifically deep learning, and how it applies to image interpretation.
- How to build deep-learning models for computer vision and how to compile bird image
- About the use of pretrained models, especially CLIP, which alone is capable of out-of-the-box bird detection with high accuracy.
- Tailoring CLIP-embedding models with small datasets for specific classification tasks.
- How to create models that go beyond classification to localisation.
- How to classify bird audio recordings.
- How to use open source tools like Merlin and BirdNet to augment research-question-specific models.

This ground-breaking volume adopts an approach based on exploring existing birding tools using AI, leading to an overview of artificial intelligence that will help build intuition about how it works. This provides a foundation for the example projects that follow, enhancing the reader's confidence in their ability to engage and participate in research involving AI. The projects are designed to guide the reader through the model-building process from dataset creation to training, testing and deployment – whether this be for image recognition, classification of calls or other new frontiers birding.

Customer Reviews

Biography

Ronald T. Kneusel has been working with machine learning in industry since 2003 and completed a PhD in artificial intelligence at the University of Colorado, Boulder, in 2016. Ron lives in Colorado, which is a great place to foster his interest in birding. Ron’s other AI books include: How AI Works (2023), Practical Deep Learning (2nd edn, 2024), and Math for Deep Learning (2021).