Deep learning has made a huge impact on a wide variety of computer vision applications. But while the capabilities of deep neural networks are impressive, understanding how to best apply them is not straightforward. In this talk, we highlight key questions that must be answered when considering incorporating a deep neural network into a vision application. What type of data will be most beneficial for the task? Should the DNN use other types of data in addition to images? How should the data be annotated? What classes should be defined? What is the minimum amount of data needed for the network to be generalized and robust? What algorithmic approach should we use for our task (classification, regression or segmentation)? What type of network should we choose (FCN, DCNN, RNN, GAN)? We’ll explain the options and trade-offs, and map out a process for making good choices for a specific application.
Applying the Right Deep Learning Model with the Right Data for Your Application
Session Speakers
-
Hila Blecher-Segev
Computer Vision and AI Research Associate, Vision Elements
Hila Blecher Segev is an expert in computer vision and AI. Her expertise includes extensive knowledge of deep learning, machine learning, 3D geometry, detection, segmentation, and other classical computer vision and image processing methods. Hila has more than 12 years’ experience developing algorithms both in academia and industry. She has MSc and BSc degrees in Engineering with honors from Ben Gurion University, Israel. Her past research focused on anomaly detection in hyperspectral data and she was part of the NASA-affiliated Earth and Planetary Image Facility. Hila also has worked in the semiconductor and defense industries. As a professional computer vision algorithms developer in a leading defense company, she received two excellence prizes for successful platform field tests. After relocating to the USA, she joined an innovative VR/AR startup. Her role focused on R&D of 3D mesh tracking with emphasis on GPU optimization.
