Рет қаралды 3,086
Endava Synthetics presents its workflow for producing synthetic training data with an end-to-end pipeline built on top of Houdini. By leveraging Houdini’s core functionality and incorporating machine learning concepts, they produce synthetic datasets used to train machine vision algorithms to detect fillings and cavities in dental x-ray imagery - involving different x-ray types like bite wing and periapical, generating dental anatomy and defects, using synthetic data to aid routine analysis, freeing up dentist time for more complex tasks.
00:00 Presentation
01:03 About Us
02:24 Understanding Machine Learning
07:20 Synthetic Data
07:48 Synthetics: Ground Truth
09:44 Synthetics: Precision
10:33 Synthetics: Variability
11:49 Synthetics: Distribution
12:57 Training Workflows
15:38 Dental X-Ray Analysis
17:17 Types of X-Rays
17:59 Types of Defects
20:15 Value to Industry
21:23 Pipeline Overview
27:44 X-Ray Sensor Proceduralism
28:37 Camera Proceduralism
29:11 Labeling
30:20 PDG Workflow Template
35:46 Testing on Synthetic Data
36:30 Understanding the Metrics
37:42 First Testing on Real Data
38:31 The Domain Gap
50:16 Q&A
Jon Hanzelka is a Technical Art Director at Endava where he oversees the synthetic pipeline development and client engagement that enables machine learning solutions across a broad range of industry verticals. He has worked in tech and entertainment for over 18 years; contributing to films such as The Avengers, Inception, and The Dark Knight Rises and helping ship products like Microsoft HoloLens v1 and 2. For the past 8 years he has been focused on the field of synthetic data and how it can be used to solve many of the industry’s most challenging machine learning problems.
Jacob Berrier is a Senior Technical Artist at Endava. His work focuses on the generation of procedural systems and the creation of pipeline tooling to produce synthetic training data for machine learning applications. Previously, Jacob started in the motion graphics industry with a collection of freelance artists who formed the company Good Secrets. He then worked at Microsoft as a Technical Artist before joining Endava. In his spare time, he restores old film cameras and prints photographs in his darkroom.