The field of 3D reconstruction and tracking is rapidly advancing, with a focus on improving accuracy and efficiency in various applications such as industrial inspection, surgical vision, and astrophotography. Recent developments have led to the creation of novel frameworks and algorithms that can handle complex scenarios, including unknown camera poses, multi-zoom image sets, and feature-deficient conditions. These innovations have the potential to revolutionize fields such as robotics, healthcare, and astronomy. Noteworthy papers in this area include: MZEN, which proposes a multi-zoom enhanced NeRF framework for 3D reconstruction with unknown camera poses, achieving significant improvements in PSNR, SSIM, and LPIPS. Surg-InvNeRF, which presents an invertible NeRF architecture for 3D tracking and reconstruction in surgical vision, surpassing state-of-the-art methods in 2D and 3D point tracking.
Advances in 3D Reconstruction and Tracking
Sources
MZEN: Multi-Zoom Enhanced NeRF for 3-D Reconstruction with Unknown Camera Poses
Neural Field Representations of Mobile Computational Photography
Benchmarking Deep Learning-Based Object Detection Models on Feature Deficient Astrophotography Imagery Dataset
Neural Beam Field for Spatial Beam RSRP Prediction
Tracking Any Point Methods for Markerless 3D Tissue Tracking in Endoscopic Stereo Images
TrackOR: Towards Personalized Intelligent Operating Rooms Through Robust Tracking
A Robust Epipolar-Domain Regularization Algorithm for Light Field Depth Estimation
Surg-InvNeRF: Invertible NeRF for 3D tracking and reconstruction in surgical vision
AR Surgical Navigation With Surface Tracing: Comparing In-SitVisualization with Tool-Tracking Guidance for Neurosurgical Applications