I’m in the process of characterizing the ZED2 depth estimation. A question that I have is for Ultra mode, is AI involved at all? Is the process of generating the depth map in Ultra mode based on non AI software and is deterministic? I would like to make sure that depth map with Ultra mode is not generated by AI and can’t really be characterized for all cases.
Please let me know if what I stated about the depth generation process is wrong and please let me know what is the technology used.
Hi @TL-4319
Welcome to the Stereolabs community.
The only depth mode that uses AI to generate the Depth Map is NEURAL.
All the other modes are based on “classic” stereo matching algorithms.
Can you go into some details about what algorithms are used in Ultra, Performance and Quality mode? Thanks
I’m sorry, but we cannot disclose this closed-source information. Our algorithms are the result of over 10 years of R&D on stereo vision.
I understand. Thanks for the respond and the great product!