Weighted point cloud augmentation for neural network training data class-imbalance
We develop a novel solution to apply a weighted augmentation to physically decrease the class-imbalance.
![Weighted point cloud augmentation for neural network training data class-imbalance Point cloud of street scene with semantic labels](https://www.ucl.ac.uk/civil-environmental-geomatic-engineering/sites/civil_environmental_geomatic_engineering/files/research-project/weighted_point_cloud_augmentation_for_neural_network_training_data_class-imbalance.jpg)
6 June 2019
A key issue when training deep neural networks for outdoor point clouds is the inevitable large data imbalance. For example, a typical street scene will contain orders of magnitudes more ground points than street furniture. We develop a novel solution to apply a weighted augmentation to physically decrease the class-imbalance.
People
- David Griffiths
- Prof. Jan Boehm