Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Concept Videos

Modeling and Similitude01:12

Modeling and Similitude

369
Scaled modeling is a fundamental technique in engineering, enabling the study of large and complex systems by creating smaller, manageable replicas that recreate critical characteristics of the original. In hydrology and civil infrastructure, for example, scaled models of dams help analyze water flow, turbulence, and pressure. This method allows for accurate predictions of real-world behavior within a controlled environment, significantly reducing the cost and time involved in full-scale...
369
Elastic Collisions: Case Study01:15

Elastic Collisions: Case Study

14.6K
Elastic collision of a system demands conservation of both momentum and kinetic energy. To solve problems involving one-dimensional elastic collisions between two objects, the equations for conservation of momentum and conservation of internal kinetic energy can be used. For the two objects, the sum of momentum before the collision equals the total momentum after the collision. An elastic collision conserves internal kinetic energy, and so the sum of kinetic energies before the collision equals...
14.6K

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Combining Synthetic Images and Deep Active Learning: Data-Efficient Training of an Industrial Object Detection Model.

Journal of imaging·2024
See all related articles

Related Experiment Video

Updated: Oct 10, 2025

Photorealistic Learned Landscapes for Augmented Reality
06:54

Photorealistic Learned Landscapes for Augmented Reality

Published on: June 27, 2025

267

Generating Images with Physics-Based Rendering for an Industrial Object Detection Task: Realism versus Domain

Leon Eversberg1, Jens Lambrecht1

  • 1Chair Industry Grade Networks and Clouds, Technische Universität Berlin, Straße des 17. Juni 135, 10623 Berlin, Germany.

Sensors (Basel, Switzerland)
|December 10, 2021
PubMed
Summary

Generating synthetic data for deep learning improves industrial object detection. Domain randomization is effective, but incorporating domain knowledge enhances performance, offering practical guidelines and tools for creating realistic training images.

Keywords:
data-centric AIdeep learningdomain randomizationimage synthesisobject detectionphysics-based renderingsynthetic images

More Related Videos

Localizing Protein in 3D Neural Stem Cell Culture: a Hybrid Visualization Methodology
21:47

Localizing Protein in 3D Neural Stem Cell Culture: a Hybrid Visualization Methodology

Published on: December 19, 2010

12.9K
Author Spotlight: Development of an Automated Camera-Based System for Real-Time Blast Overpressure Monitoring and TBI Risk Assessment in Military Training
06:20

Author Spotlight: Development of an Automated Camera-Based System for Real-Time Blast Overpressure Monitoring and TBI Risk Assessment in Military Training

Published on: December 6, 2024

3.0K

Related Experiment Videos

Last Updated: Oct 10, 2025

Photorealistic Learned Landscapes for Augmented Reality
06:54

Photorealistic Learned Landscapes for Augmented Reality

Published on: June 27, 2025

267
Localizing Protein in 3D Neural Stem Cell Culture: a Hybrid Visualization Methodology
21:47

Localizing Protein in 3D Neural Stem Cell Culture: a Hybrid Visualization Methodology

Published on: December 19, 2010

12.9K
Author Spotlight: Development of an Automated Camera-Based System for Real-Time Blast Overpressure Monitoring and TBI Risk Assessment in Military Training
06:20

Author Spotlight: Development of an Automated Camera-Based System for Real-Time Blast Overpressure Monitoring and TBI Risk Assessment in Military Training

Published on: December 6, 2024

3.0K

Area of Science:

  • Computer Vision
  • Deep Learning
  • Machine Learning

Background:

  • Limited training data hinders industrial deep learning applications.
  • Synthetic image generation is a potential solution, but the domain gap between synthetic and real images is a challenge.

Purpose of the Study:

  • To explore physics-based rendering for generating synthetic images in industrial object detection.
  • To compare domain randomization versus domain knowledge strategies for minimizing the domain gap.

Main Methods:

  • Investigated various rendering parameters: lighting, background, object texture, and foreground objects.
  • Employed a data-centric approach comparing different realism and variability levels.
  • Evaluated performance using average precision metrics.

Main Results:

  • Domain randomization proved to be a viable strategy for industrial object detection.
  • Utilizing domain knowledge for object-specific aspects improved detection performance.
  • Different strategies yielded varying levels of detection accuracy.

Conclusions:

  • Domain randomization is effective for industrial object detection using synthetic data.
  • Domain knowledge can be strategically applied to further enhance detection.
  • Provided practical guidelines and an open-source tool for synthetic image generation in industry.