Point Ladder Tuning: Parameter-Efficient Hierarchical Adaptation for 3D Point Cloud Understanding

Junlin Chang, Longhao Zou, Rui Li†

Geometry-Adaptive 3D Convex Polyhedron for Tomographic Reconstruction

Yaojian Xu, Rui Li†, Zhiye Tang, Longhao Zou, Xu Wang

CWC-DNERF: Compact Dynamic Neural Radiance Field VIA Discrete Wavelet Transform And Learnable Codebooks

Yaojian Xu, Rui Li†, Q. Zhang, L. Zou, Q. Liu, Xu Wang

Enhancing telecooperation through haptic twin for internet of robotic things: Implementation and challenges

M. Huang, R. Feng, L. Zou, Rui Li, J. Xie

Adaptive differentiable grids for cryo-electron tomography reconstruction and denoising

Y. Wang, R. Idoughi, D. Rückert, Rui Li, W. Heidrich

NeAT: Neural Adaptive Tomography

Darius Rückert, Yuanhao Wang, Rui Li, Ramzi Idoughi, Wolfgang Heidrich

NeAT is a neural adaptive tomography method that reconstructs 3D volumes from sparse and limited-angle CT projections using a learned, adaptive sampling strategy within a differentiable rendering framework.

Neural Adaptive Scene Tracing

Rui Li*, Darius Rückert, Yuanhao Wang, Ramzi Idoughi, Wolfgang Heidrich

Shape and Reflectance Reconstruction in Uncontrolled Environments by Differentiable Rendering

Rui Li*, Guangmin Zang, Miao Qi, Wolfgang Heidrich

Simultaneous reconstruction of geometry and reflectance in uncontrolled environments from multi-view photography using hand-held cameras. Builds a virtual scene in a differentiable rendering system, optimized by alternating and stochastic photometric objectives, generating photo-realistic novel views. Superior to SOTA in novel view synthesis.

IntraTomo: Self-supervised Learning-based Tomography via Sinogram Synthesis and Prediction

Guangming Zang, Ramzi Idoughi, Rui Li, Peter Wonka, Wolfgang Heidrich

Combines learning-based and model-based approaches for ill-posed CT inverse problems. Two modules: sinogram prediction (density field as continuous differentiable NN function, self-supervised from incomplete/degraded sinogram) and geometry refinement (local & non-local geometrical priors), applied iteratively. Outperforms on limited-angle tomography (45°), sparse view (as few as 8 views), super-resolution (8×).

Reflection Separation via Multi-bounce Polarization State Tracing

Rui Li*, Simeng Qiu*, Guangming Zang, Wolfgang Heidrich

Generalizes reflection removal to real-world complex light interactions. Learning framework for supervised reflection separation with a polarization-guided ray-tracing model. Uses a polarization sensor capturing 4 linearly polarized photos simultaneously. A new polarization-guided image formation model plus supervised learning for the ray-tracing model yields unprecedented reconstruction quality on real and synthetic data. († equal contribution)

Hierarchical and View-invariant Light Field Segmentation by Maximizing Entropy Rate on 4D Ray Graphs

Rui Li*, Wolfgang Heidrich

A new light field segmentation method respecting texture appearance, depth consistency, and occlusion. Creates well-shaped segments robust to viewpoint changes; hierarchical — a single optimization yields a whole hierarchy of segmentations. Uses a submodular objective function optimized greedily; introduces a "disjoint tree" data structure for efficient submodular optimization on very large graphs.

Monocular Long-term Target Following on UAVs

Rui Li*, Minjian Pang, Cong Zhao, Guyue Zhou, Lu Fang

Long-term visual tracking on UAVs. Exploits correlation between a frequency tracker and a spatial detector; novel FAST algorithm. Robustness (frequency tracker → spatial detector covers temporal variance/invariance) plus efficiency (coarse-to-fine redetection, no extra classifier / exhaustive search). Implemented on a quadrotor for indoor/outdoor real-time automatic smooth long-term target following.

Cluster Sensing Superpixel and Grouping

Rui Li*, Lu Fang

Cluster Sensing Superpixel (CSS) method. Cluster centers have representativeness (local max pixel density) and isolation; CSS identifies centers via pixel density. Integrates superpixel cues into a bipartite graph segmentation framework, applied to microscopy image segmentation. ~5× faster than SOTA with comparable performance.