US Patent Application 17827201. Using Histograms For Self-Supervised Depth Estimation simplified abstract
Contents
Using Histograms For Self-Supervised Depth Estimation
Organization Name
TOYOTA JIDOSHA KABUSHIKI KAISHA
Inventor(s)
Vitor Guizilini of Santa Clara CA (US)
Rares A. Ambrus of San Francisco CA (US)
Sergey Zakharov of San Francisco CA (US)
Using Histograms For Self-Supervised Depth Estimation - A simplified explanation of the abstract
This abstract first appeared for US patent application 17827201 titled 'Using Histograms For Self-Supervised Depth Estimation
Simplified Explanation
- The patent application describes an improved approach to training a depth model to estimate depth from monocular images. - The method involves determining loss values based on a photometric loss function, which are associated with a depth map derived from a monocular image. - Histograms are generated for the loss values corresponding to different regions of a target image. - Erroneous values of the loss values are identified and masked to avoid considering them during training of the depth model. - This approach helps improve the accuracy and reliability of depth estimation from monocular images. - The use of histograms allows for better analysis and identification of erroneous values. - By masking the erroneous values, the training process can focus on more reliable data, leading to a more effective depth model.
Original Abstract Submitted
System, methods, and other embodiments described herein relate to an improved approach to training a depth model to derive depth estimates from monocular images using histograms to assess photometric losses. In one embodiment, a method includes determining loss values according to a photometric loss function. The loss values are associated with a depth map derived from an input image that is a monocular image. The method includes generating histograms for the loss values corresponding to different regions of a target image. The method includes, responsive to identifying erroneous values of the loss values, masking the erroneous values to avoid considering the erroneous values during training of the depth model.
