SeqNet: Learning Descriptors for Hierarchical Place Recognition | Sourav Garg, PostDoc @QUT

Paper Abstract:
Visual Place Recognition (VPR) is key to Visual Localization and Autonomous Robot Navigation. Robots move and accrue information sequentially. Existing sequential VPR methods are inaccurate, induce high latency, and perform compute-intensive dense retrieval. We propose SeqNet and sequence-based Hierarchical VPR as an accurate, low latency, and efficient approach to visual localization, achieving state-of-the-art results on standard public benchmarks.
Speaker Bio:
Dr. Sourav Garg is currently leading the Perception and Localization research theme at the QUT Centre for Robotics (QCR) as a Postdoctoral Research Fellow. Sourav is a robotic-vision enthusiast. His research spans computer vision, robotics, and deep learning, motivated by practical applications that involve a moving camera. He pioneered research in the twin challenges of visual place recognition that requires dealing with scene appearance and camera viewpoint simultaneously. His award-winning research and Ph.D. thesis proposed novel ways of robot localization based on visual semantics, inspired by humans. He is always keen on exploring research problems related to scene understanding and robot navigation, particularly those revolving around effective representation and matching of visual information.
Paper Link: arxiv.org/abs/2102.11603