Self-supervised Visual Reinforcement Learning with Object-centric Representations.
Andrii ZadaianchukMaximilian SeitzerGeorg MartiusPublished in: CoRR (2020)
Keyphrases
- reinforcement learning
- object description
- visual objects
- object models
- visual information
- visual appearance
- object representation
- mid level
- visual input
- state space
- visual features
- multiple objects
- visual representations
- object representations
- multi agent
- spatial relations
- object segmentation
- high level
- visual patterns
- visual properties
- visual cues
- complex objects
- data objects
- low level
- d objects
- object model
- optimal policy
- machine learning
- moving objects
- viewpoint
- learning process
- real world objects
- higher level
- keypoints
- object tracking
- representation scheme
- visual perception
- visual data
- function approximation