Synthetic Data for Computer Vision Problems When Real Annotation Is Too Expensive

Explore the potential of synthetic data generation for computer vision problems in this 26-minute conference talk by Abgeiba Isunza Navarro at the 2023 GAIA Conference. Delve into the challenges of obtaining large amounts of labeled data for deep learning models and discover how synthetic data can serve as a viable alternative. Learn about the latest developments in synthetic data generation, including graphic tools, simulator engines, and generative models. Gain insights into building effective computer vision models primarily using synthetic data, with practical examples of use cases and tools. Understand how to approach computer vision tasks when faced with limited labeled data. Benefit from the speaker's expertise as a machine learning engineer at Modulai, drawing from her experience in developing AI products across various industries.

Synthetic Data for Computer Vision Problems, When Real Annotation Is Too Expensive by Abgeiba I. N.