17839201. SYSTEM AND METHOD FOR 3D OBJECT PERCEPTION TRAINED FROM PURE SYNTHETIC STEREO DATA simplified abstract (TOYOTA JIDOSHA KABUSHIKI KAISHA)

From WikiPatents
Jump to navigation Jump to search

SYSTEM AND METHOD FOR 3D OBJECT PERCEPTION TRAINED FROM PURE SYNTHETIC STEREO DATA

Organization Name

TOYOTA JIDOSHA KABUSHIKI KAISHA

Inventor(s)

Thomas Kollar of San Jose CA (US)

Kevin Stone of Palo Alto CA (US)

Michael Laskey of Oakland CA (US)

Mark Edward Tjersland of Mountain View CA (US)

SYSTEM AND METHOD FOR 3D OBJECT PERCEPTION TRAINED FROM PURE SYNTHETIC STEREO DATA - A simplified explanation of the abstract

This abstract first appeared for US patent application 17839201 titled 'SYSTEM AND METHOD FOR 3D OBJECT PERCEPTION TRAINED FROM PURE SYNTHETIC STEREO DATA

Simplified Explanation

The patent application describes a method for perceiving 3D objects using synthetic stereo images and neural networks. Here are the key points:

  • The method involves extracting features from each image of a synthetic stereo pair of images.
  • A low-resolution disparity image is generated based on the extracted features from each image of the stereo pair.
  • A trained neural network predicts a feature map using the low-resolution disparity image and one of the stereo pair images.
  • A perception prediction head generates a perception prediction of a detected 3D object based on the predicted feature map.

Potential Applications

This technology has potential applications in various fields, including:

  • Autonomous driving: Enhancing object detection and perception for self-driving vehicles.
  • Robotics: Improving the ability of robots to perceive and interact with their environment.
  • Augmented reality: Enhancing the realism and accuracy of virtual objects in augmented reality applications.
  • Medical imaging: Assisting in the analysis and interpretation of 3D medical images.

Problems Solved

The method addresses the following problems:

  • Accurate 3D object perception: By utilizing synthetic stereo images and neural networks, the method improves the accuracy of perceiving 3D objects.
  • Low-resolution disparity estimation: The method generates a low-resolution disparity image, which helps in estimating the depth information of the objects.
  • Efficient feature extraction: Extracting features from the synthetic stereo pair of images allows for efficient and effective object perception.

Benefits

The use of this technology offers several benefits:

  • Improved object detection: The method enhances the ability to detect and recognize 3D objects in various scenarios.
  • Real-time perception: By utilizing neural networks, the method enables real-time perception of 3D objects.
  • Enhanced depth estimation: The low-resolution disparity image aids in estimating the depth information of objects accurately.
  • Versatile applications: The technology can be applied in multiple domains, including autonomous driving, robotics, augmented reality, and medical imaging.


Original Abstract Submitted

A method for 3D object perception is described. The method includes extracting features from each image of a synthetic stereo pair of images. The method also includes generating a low-resolution disparity image based on the features extracted from each image of the synthetic stereo pair images. The method further includes predicting, by a trained neural network, a feature map based on the low-resolution disparity image and one of the synthetic stereo pair of images. The method also includes generating, by a perception prediction head, a perception prediction of a detected 3D object based on the feature map predicted by the trained neural network.

Retrieved from "http://wikipatents.org/index.php?title=17839201._SYSTEM_AND_METHOD_FOR_3D_OBJECT_PERCEPTION_TRAINED_FROM_PURE_SYNTHETIC_STEREO_DATA_simplified_abstract_(TOYOTA_JIDOSHA_KABUSHIKI_KAISHA)&oldid=17452"