Vision Paper
Vision research currently focuses on developing robust and efficient methods for processing and understanding visual information, often integrating it with other modalities like language and touch. Key areas include improving the accuracy and efficiency of models like transformers and exploring alternatives such as Mamba and structured state space models for various tasks, ranging from object detection and segmentation to navigation and scene understanding. This work is driven by the need for improved performance in applications such as robotics, autonomous systems, medical image analysis, and assistive technologies, with a strong emphasis on addressing challenges like limited data, computational cost, and generalization to unseen scenarios.
Papers
See, Think, Confirm: Interactive Prompting Between Vision and Language Models for Knowledge-based Visual Reasoning
Zhenfang Chen, Qinhong Zhou, Yikang Shen, Yining Hong, Hao Zhang, Chuang Gan
Toward Building General Foundation Models for Language, Vision, and Vision-Language Understanding Tasks
Xinsong Zhang, Yan Zeng, Jipeng Zhang, Hang Li
Fully-attentive and interpretable: vision and video vision transformers for pain detection
Giacomo Fiorentini, Itir Onal Ertugrul, Albert Ali Salah
Learning on the Job: Self-Rewarding Offline-to-Online Finetuning for Industrial Insertion of Novel Connectors from Vision
Ashvin Nair, Brian Zhu, Gokul Narayanan, Eugen Solowjow, Sergey Levine