Title:	Automatic Annotations for Autonomous Driving
DNr:	Berzelius-2023-174
Project Type:	LiU Berzelius
Principal Investigator:	Adam Tonderski <adam.tonderski@zenseact.com>
Affiliation:	Lunds universitet
Duration:	2023-06-21 – 2024-01-01
Classification:	10207
Keywords:

Abstract

In this project we investigate develop various approaches for generating automatic annotations for autonomous driving applications. This ranges from: - ensembles of powerful offline detectors - tracking/smoothing, using the future to improve predictions of the present - synthetic data generation - neural reconstruction for semi-synthetic data generation (object manipulation, novel view generation, etc.)