Accurate detection and 3D localization of humans using a novel YOLO-based RGB-D fusion approach and synthetic training data.
Timm LinderKilian Y. PfeifferNarunas VaskeviciusRobert SchirmerKai Oliver ArrasPublished in: ICRA (2020)
Keyphrases
- training data
- accurate localization
- multi modal fusion
- real world
- activity detection
- data sets
- detection method
- completely automated
- training set
- false positives
- automatic detection
- multi sensor
- manually labeled training data
- decision level fusion
- high quality
- accurate registration
- human behavior
- data fusion
- high accuracy
- object detection
- mobile robot
- highly accurate
- image fusion
- object localization
- noisy data
- test data
- computationally efficient
- learned from training data
- precise localization
- classification accuracy