System and method for generating synthetic training data
AssigneesDENSO International America, Inc., Denso Corporation
InventorShawn Hunt
Life of this patent
Estimating the term…
Abstract
A system for generating synthetic training data may include one or processors and a memory in communication with the one or more processors and having a receiving module, a duplication module, and an insertion module. The modules have instructions that when executed by the one or more processors cause the one or more processors to receive original training data being in the form of a three-dimensional point cloud and having one or more original objects formed by at least a portion of the three-dimensional point cloud and annotated with original annotation data, duplicate one of the one or more original objects to generate a synthetic object, and insert the synthetic object within the original training data to generate the synthetic training data.
Claims20
A system for generating synthetic training data, the system comprising:
- one or processors;
- and
- a memory in communication with the one or more processors, the memory having:
- a receiving module, the receiving module having instructions that, when executed by the one or more processors, cause the one or more processors to receive original training data, the original training data being in the form of a three-dimensional point cloud generated by a sensor and having one or more original objects formed by at least a portion of the three-dimensional point cloud generated by the sensor, the one or more original objects being annotated with original annotation data,
- a duplication module, the duplication module having instructions that, when executed by the one or more processors, cause the one or more processors to duplicate one of the one or more original objects to generate a synthetic object, the synthetic object being formed of duplicate portions of the three-dimensional point cloud generated by the sensor that formed the one of the one or more original objects and being annotated with synthetic annotated data, the synthetic annotated data being a duplication of the original annotation data of the one or more original objects, and
- an insertion module, the insertion module having instructions that, when executed by the one or more processors, cause the one or more processors to insert the synthetic object within the original training data to generate the synthetic training data.
A method for generating synthetic training data, the method comprising the steps of:
- receiving original training data, the original training data being in the form of a three-dimensional point cloud generated by a sensor and having one or more original objects formed by at least a portion of the three-dimensional point cloud generated by the sensor, the one or more original objects being annotated with original annotation data;
- duplicating the one or more original objects to generate a synthetic object, the synthetic object being formed of duplicate portions of the three-dimensional point cloud generated by the sensor that formed the one of the one or more original objects and being annotated with synthetic annotated data, the synthetic annotated data being a duplication of the original annotation data of the one or more original objects;
- and
- inserting the synthetic object within the original training data to generate the synthetic training data.
A non-transitory computer-readable medium storing instructions for generating synthetic training data that, when executed by one or more processors, cause the one or more processors to:
- receive original training data, the original training data being in the form of a three-dimensional point cloud generated by a sensor and having one or more original objects formed by at least a portion of the three-dimensional point cloud generated by the sensor, the one or more original objects being annotated with original annotation data;
- duplicate one of the one or more original objects to generate a synthetic object, the synthetic object being formed of duplicate portions of the three-dimensional point cloud generated by the sensor that formed the one of the one or more original objects and being annotated with synthetic annotated data, the synthetic annotated data being a duplication of the original annotation data of the one or more original objects;
- and
- insert the synthetic object within the original training data to generate the synthetic training data.
Dependencies are inferred from the claim text.