Nerf
PulseAugur coverage of Nerf — every cluster mentioning Nerf across labs, papers, and developer communities, ranked by signal.
6 day(s) with sentiment data
-
New RLHF method fine-tunes 3D GANs directly from human preferences
Researchers have developed a novel method for fine-tuning 3D-aware generative models, specifically a face GAN called EG3D, using reinforcement learning from human feedback (RLHF). This approach directly optimizes the ne…
-
NeRF-based 3D detector improves autonomous driving perception
Researchers have developed a novel NeRF-Resembled Point-based 3D detector (NeRP3D) that addresses limitations in current NeRF-based pre-training for autonomous driving. Existing methods force NeRFs to work with view tra…
-
New dataset enables novel view synthesis for gastroendoscopy
Researchers have introduced the GastroNVS dataset, the first real-world dataset designed for novel view synthesis (NVS) in gastroendoscopy. This dataset aims to address the current lack of sufficient data for evaluating…
-
Render-FM achieves real-time photorealistic CT scan rendering
Researchers have developed Render-FM, a novel feedforward model designed for real-time photorealistic volumetric rendering of CT scans. This model significantly speeds up the rendering process, reducing it from hours or…
-
VisDom improves sparse novel view synthesis with visible domain constraint · 2 sources tracked
Researchers have introduced VisDom, a novel geometric constraint for sparse novel view synthesis that improves 3D reconstruction quality from limited input images. This learning-free method enhances existing NeRF and Ga…
-
EDoF-NeRF enhances depth-of-field for photorealistic neural radiance fields
Researchers have developed EDoF-NeRF, a novel method to enhance the depth-of-field (DoF) in neural radiance fields (NeRF) for more photorealistic novel view rendering. This technique utilizes a coded aperture placed at …
-
Shanghai Jiao Tong U. advances SLAM for dynamic, deformable environments
Researchers at Shanghai Jiao Tong University, led by Professor Hesheng Wang, are advancing Simultaneous Localization and Mapping (SLAM) beyond static environments. Their work focuses on enabling robots to navigate and u…
-
DeblurNVS framework synthesizes novel views from motion-blurred images
Researchers have developed DeblurNVS, a new framework designed to synthesize high-fidelity novel views from sparse, motion-blurred images without needing per-scene optimization. This method works by restoring intermedia…
-
NeRF methods boost spacecraft pose estimation and 3D reconstruction
Researchers have developed new methods using Neural Radiance Fields (NeRF) to improve spacecraft pose estimation and 3D reconstruction from imagery. One approach uses NeRF-based augmentations to train pose estimators wi…
-
Diffusion models and NeRF combine for probabilistic 3D scene reconstruction
Researchers have developed a novel method for 3D scene reconstruction by integrating diffusion models with Neural Radiance Fields (NeRF). This approach treats 3D reconstruction as a probabilistic problem, using a stocha…
-
PropSplat method reconstructs RF fields without 3D maps
Researchers have developed PropSplat, a novel method for reconstructing radio frequency (RF) fields without relying on detailed 3D maps or extensive measurement campaigns. This approach utilizes 3D anisotropic Gaussian …
-
MapRF uses NeRF-guided self-training for weakly supervised HD map construction
Researchers have developed MapRF, a novel framework for constructing high-definition (HD) maps for autonomous driving systems using only 2D image labels. This weakly supervised approach leverages Neural Radiance Fields …
-
Geometry-aware strategies improve novel view synthesis for indoor scenes
Researchers have developed new methods for efficiently allocating representation capacity in neural radiance fields (NeRFs) to generate immersive indoor environments. By leveraging geometric priors from estimated scene …
-
SaLF offers real-time multi-sensor simulation for autonomous driving
Researchers have developed SaLF, a novel volumetric representation for real-time multi-sensor simulation. This method uses a sparse set of 3D voxel primitives, each containing a local implicit field, to achieve faster t…
-
PanDORA system captures high dynamic range radiance for realistic image-based lighting
Researchers have developed PanDORA, a system for rapidly capturing high dynamic range (HDR) radiance maps essential for realistic image-based lighting. Unlike existing methods that rely on slow exposure bracketing or lo…
-
HumanSplatHMR refines 3D human pose and avatar generation via joint optimization
Researchers have introduced HumanSplatHMR, a novel framework that jointly optimizes 3D human pose estimation and avatar creation from video. This approach addresses limitations in existing methods by integrating pose re…
-
GOR-IS framework improves 3D object removal with intrinsic space inpainting
Researchers have developed GOR-IS, a new framework for removing objects from 3D scene reconstructions generated by methods like 3D Gaussian Splatting. This approach addresses limitations in existing techniques by explic…
-
New EAG-PT method reconstructs and edits indoor scenes with diffuse global illumination
Researchers have developed EAG-PT, a novel method for reconstructing and editing indoor scenes with diffuse global illumination. This approach utilizes 2D Gaussians as a geometric proxy, bypassing the need for explicit …
-
RayFormer AI model achieves SOTA in video snapshot compressive imaging
Researchers have introduced RayFormer, a novel approach for video snapshot compressive imaging (SCI) that improves reconstruction quality. This method utilizes a patch-level ray sampling strategy and an Inter- and Intra…
-
Researchers unify visual and wireless sensing for 3D radio map construction
Researchers have developed a novel framework called URF-GS that unifies visual and wireless sensing for constructing 3D radio maps. This approach leverages 3D Gaussian splatting and inverse rendering to integrate cross-…