Utility Functions Overview

OnSite provides a set of Utility Functions for learning autonomous-driving problems and tasks. These dataset- and model-oriented guides and high-level interfaces help learners get started quickly with research and development.

  • Environment Setup

  • Data Filtering

  • Data Split

  • Model Evaluation

Environment Setup Guide

For selected datasets on the OnSite Learning Center, download links and guides are provided. Beginners can run the given commands step by step to complete Environment Setup, data import, and other prerequisites.

Data Filtering

Autonomous-driving datasets contain data from many scenarios and factors. Data Filtering lets researchers combine scenario, feature, and operating-condition filters to create customized datasets.

Data Split

Because datasets vary in size, several data-splitting functions are provided to help beginners quickly create training and test sets while preserving realistic data distributions.

Model Evaluation

For different autonomous-driving tasks, a set of functions is provided to calculate model-performance metrics commonly used in research. Visualization functions are also available for selected tasks.