Perception Model
Perception models aim to enable machines to understand and interpret sensory information, mirroring human perception. Current research emphasizes improving model robustness and efficiency through techniques like data augmentation with physics-based simulations, uncertainty quantification for resource-efficient foundation model refinement, and the integration of multimodal data (e.g., visual and textual information) using architectures such as transformers and generative models. These advancements are crucial for enhancing the reliability and safety of applications like autonomous driving, robotics, and medical imaging, where accurate and robust perception is paramount.
Papers
Uni-Perceiver: Pre-training Unified Architecture for Generic Perception for Zero-shot and Few-shot Tasks
Xizhou Zhu, Jinguo Zhu, Hao Li, Xiaoshi Wu, Xiaogang Wang, Hongsheng Li, Xiaohua Wang, Jifeng Dai
SEAL: Self-supervised Embodied Active Learning using Exploration and 3D Consistency
Devendra Singh Chaplot, Murtaza Dalal, Saurabh Gupta, Jitendra Malik, Ruslan Salakhutdinov