The Importance of Ray Tracing by Stanford University

The Importance of Ray Tracing by Stanford University

Stanford-University-Ray-Tracing-Article

Check out this interesting read by Stanford University on the importance of ray tracing and camera parameters for the generation of synthetic data used in machine learning! 

Variations in specific camera parameters and image processing operations impact convolutional neural network (CNN) generalization. 

The researchers come to the following conclusion:

The generalization between the ray-traced synthetic imagesand the camera images is encouraging. The performancelevel using synthetic images for the car detection applicationis already adequate to be helpful in advancing all threeobjectives (universal network design, specific camera sim-ulation, co-design of camera and network).

At ANYVERSE we place great focus on camera/sensor parameters for each dataset generated. We can customize each sensor model by using our proprietary unbiased spectral render engine. So if you are looking for a specific lens, we have it and we can tweak camera features and scenario specifics to match your exact model. Give it a try by designing your own dataset!

Scroll to Top

Let's talk synthetic data!

Design your dataset

Let us take you through the steps required for a customizable ANYVERSE dataset. 
We will provide you with some samples afterwards or will get in touch to understand better your data needs.

1
Application
2
Scenario
3
Sensors
4
Camera Lens
5
Resolution
6
Data Output
7
About you

Let’s start with the practical use of your dataset!

Next

Where does the action take place? Pick a scenario!

Previous
Next

Would you like to add camera(s) and/or LiDAR to your ego vehicle?

Previous
Next

Do you require a specific camera lens ?

Previous
Next

What resolution does your model require?

Previous
Next

Finally, tick all the additional metadata channels you’d like to have!

Previous
Next
Send

Sign up to our newsletter!