Training of neural networks using basic truth data supplemented with map information for autonomous machine applications

摘要

In various embodiments, training sensor data generated by one or more sensors of autonomous machines can be associated with high-resolution (HD) map data to supplement and/or generate ground truth data - in embodiments, e.g. B. automatically. The ground truth data can be associated with the training sensor data to train one or more deep neural networks (DNNs) to compute outputs corresponding to autonomous machine operations - such as B. object or feature detection, road feature detection and classification, waiting condition detection and classification, etc. Consequently, the HD map data can be leveraged during training, allowing the DNNs - in use - autonomous machines in safe navigation in environments without relying solely on HD map data.