The Fifth Elephant 2023 Monsoon

The integration of computer vision technologies in the automotive industry has revolutionized various aspects of vehicle safety, navigation, and driver assistance systems. However, developing robust and accurate computer vision models for real-world scenarios necessitates large-scale, diverse, and accurately labeled datasets, which can be challenging to obtain. Consider a scenario where the duration of data collection and annotation efforts, spanning from 60 to 90 days and involving hundreds of manual annotators, is condensed to a mere 2 hours of effort resulting in impeccable, error-free data.

Drawing on our group’s experience at Mercedes-Benz R&D India in research as well as delivering customer-ready products in the area of intelligent interior and in-cabin sensing (MBUX interior assist), this talk will explore the benefits of using synthetic data to augment or replace traditional, labor-intensive data collection methods.

We will delve into the techniques and methodologies employed for generating synthetic data tailored to computer vision tasks. This includes modeling virtual environments, capturing complex actions, and incorporating accurate ground truth annotations.

Furthermore, we will examine the key challenges involved in using synthetic data that closely matches real-world data distributions and the strategies used to bridge the domain gap between synthetic and real data. This talk will highlight successful applications of synthetic data in automotive domain. Specifically, we will focus on the techniques we employed on datasets like cityscapes, achieving state-of-the-art results (Top-1 when submitted, now Top-3).