Labels Are Not Perfect: Improving Probabilistic Object Detection via Label Uncertainty.
Di FengLars RosenbaumFabian TimmKlaus DietmayerPublished in: CoRR (2020)
Keyphrases
- object detection
- uncertain data
- multi label
- class labels
- label space
- image labeling
- conditional probabilities
- label noise
- label propagation
- multi label classification
- label assignment
- labeling process
- bayesian networks
- face detection
- decision theory
- multi label learning
- probabilistic model
- probabilistic knowledge
- training data
- active learning
- probability theory
- image parsing
- generative model
- probability measures
- pairwise
- multi instance learning
- labeling problems
- semantic labels
- active learning framework
- handling uncertainty
- scene understanding
- object class
- semi supervised
- probabilistic reasoning
- partially labeled
- multilabel classification
- multiple labels
- belief functions
- inference process
- text categorization
- pedestrian detection
- computer vision
- object classification