PR-RL: Portrait Relighting Via Deep Reinforcement Learning.
Xiaoyan ZhangYukai SongZhuopeng LiJianmin JiangPublished in: IEEE Trans. Multim. (2022)
Keyphrases
- reinforcement learning
- reinforcement learning algorithms
- function approximation
- model free
- rl algorithms
- state space
- temporal difference
- optimal policy
- illumination conditions
- machine learning
- temporal difference learning
- markov decision processes
- partially observable domains
- learning problems
- dynamic programming
- multi agent
- control problems
- lighting conditions
- optimal control
- learning classifier systems
- learning process
- partially observable
- learning agents
- learning algorithm
- action space
- action selection
- real scenes
- policy iteration
- function approximators
- spherical harmonics
- autonomous learning
- approximate dynamic programming
- multi agent reinforcement learning
- policy search
- state and action spaces
- reward shaping
- face recognition
- continuous state and action spaces