Talk: DeepWhale: Recognizing and Localizing Endangered Right Whales with Extremely Deep Neural Networks

Speaker: Felix Lau

Language: Cantonese (with English Slides)

Photo of Felix Lau

About Speaker

Felix Lau is currently a machine learning scientist at Arterys, a San Francisco-based startup. He applies machine learning to medical imaging data and help clinicians to make timely and accurate diagnosis. Previously, he built a computer vision and analytics system for retailing industry using deep learning.

About the Topic

The objective of the Kaggle Right Whale of the challenge is to build a facial recognizer for endangered right whales from aerial photos. This task is extremely challenging for humans because callosity pattern that is used to identify the whales is very subtle. It is also very challenging machine learning problem because there are very few training samples per individual whale. I will explain how I finished in 2nd place by utilizing state-of-the-art deep residual neural network to localize and classify these aerial photos. See for details

Tags: Deep Learning, Kaggle