We specialized in generating synthetic underwater images for robotics and computer vision. The sea and high-tech, a perfect fit. We provide high-quality images at any resolution, based on physical-based rendering, including the required ground truth data for training computer vision models.
Synthetic underwater images are computer-generated images that simulate an underwater environment on a realistic level. They can be created for various purposes, such as visual effects in films and video games, virtual reality simulations, scientific research, or educational purposes. Synthetic underwater images can include virtual representations of underwater landscapes, sea creatures, and other underwater features, such as shipwrecks, corals, and plants.
In the field of computer vision for underwater robotics, one of the biggest challenges is to obtain good ground truth data. With our computer-generated synthetic image data, we can provide any kind of ground truth you need. An example are semantic segmentation maps like shown here to the left.
Clear open ocean water, or rather turbid coastal water? To be as realistic as possible, our simulations are based on the absorption and scattering spectra of different Jerlov water types.
Jerlov water types are a system of categorizing marine water bodies based on their spectral optical attenuation. There are 5 typical open-ocean water types (I, IA, IB, II, III), and 5 coastal water types (1C, 3C, 5C, 7C, 9C). Besides these standard water types, we can include any more specific absorption and scattering spectra most appropriate for your specific situation.
Shallow waters, or deeper Ocean zones. With a full 3D simulation and the physical-based absorption and scattering spectra we ensure that light and colors are correctly reproduced.