NICE-SLAM: Neural Implicit Scalable Encoding for SLAM

By Zihan Zhu, Songyou Peng, Viktor Larsson, Weiwei Xu, Hujun Bao, Zhaopeng Cui, Martin R. Oswald, Marc Pollefeys. Computer Vision and Pattern Recognition, 2021

1,063 citations recorded by the source corpus.

Neural implicit representations have recently shown encouraging results in various domains, including promising progress in simultaneous localization and mapping (SLAM). Nevertheless, existing methods produce over- smoothed scene reconstructions and have difficulty scaling up to large scenes. These limitations are mainly due to their simple fully-connected network architecture that does not incorporate local information in the observations. In this paper, we present NICE-SLAM, a dense SLAM system that incorporates multi-level local information by introducing a hierarchical scene representation. Optimizing this representation with pre-trained geometric priors enables detailed reconstruction on large indoor scenes. Compared to recent neural implicit SLAM systems, our approach is more scalable, efficient, and robust. Experiments on five challenging datasets demonstrate competitive results of NICE-SLAM in both mapping and tracking quality. Project page: https://pengsongyou.github.io/nice-slam.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition, Geology

View Paper: https://www.semanticscholar.org/paper/18ba8f0efb362e08903e8e35b062e2c69126e371

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Aerospace Engineering

In Computer Vision and Pattern Recognition

In Geology

Citations in the corpus

NeRF-SLAM: Real-Time Dense Monocular SLAM with Neural Radiance Fields (2022, methodology): “Unfortunately, these approaches are too slow for online inference due to their choice of a large MLP as map representation. iMap[27] and Nice-SLAM[28] subsequently showed how to build accurate 3D reconstructions, without the need for poses, by partially de-coupling pose and depth estimation, similar”
ConceptGraphs: Open-Vocabulary 3D Scene Graphs for Perception and Planning (2023, background): “Closed-vocabulary semantic mapping in 3D. Early works reconstruct the 3D map through online algorithms like simultaneous localization and mapping (SLAM) [1]–[5] or offline methods like structure-from-motion (SfM) [6], [7].”
HexPlane: A Fast Representation for Dynamic Scenes (2023, background): “NeRF [46] and its variants [2, 3, 44, 48, 74, 76, 85, 92] show impressive results on novel view synthesis [9,80,87,99] and many other applications including 3D reconstruction [42, 72, 90, 94, 100], semantic segmentation [27,59,98], generative model [5,6,10, 49,62,82], and 3D content creation [1,23,3”
NeRF: Neural Radiance Field in 3D Vision, A Comprehensive Review (2022, methodology): “Building on iMAP, NICE-SLAM [74] (December 2021) improved on various aspects such as keyframe selection and NeRF architecture.”
MonoSDF: Exploring Monocular Geometric Cues for Neural Implicit Surface Reconstruction (2022, methodology): “More specifically, we investigate the following architectures: a single, large MLP [46, 71, 75, 76], a dense SDF grid [30], a single feature grid [27, 36, 50, 51] and multi-resolution feature grids [10, 21, 42, 65, 83].”
NoPe-NeRF: Optimising Neural Radiance Field with No Pose Prior (2022, methodology): “One category of the methods [33, 38, 60] use a SLAM-style pipeline, that either requires RGB-D inputs or relies on accurate camera poses generated from the SLAM tracking system.”
A Survey on 3D Gaussian Splatting (2024, positive): “These approaches adopt a hybrid deformation strategy: linear blend skinning handles rigid skeletal transformations such as joint rotations, while pose-conditioned deformation fields account for secondary non-rigid effects like muscle jiggles.”
SplaTAM: Splat, Track & Map 3D Gaussians for Dense RGB-D SLAM (2023, methodology): “These radiance-field - based SLAM algorithms [30, 54] benefit from high-fidelity global maps and image reconstruction losses that capture dense photometric information via differentiable rendering.”
Recent advances in 3D Gaussian splatting (2024, positive): “…[119] initializes Gaussians with LiDAR point cloud and builds up an optimizable size-adaptive voxel Methods PSNR ↑ SSIM ↑ LPIPS ↓ NICE-SLAM [122] 24.42 0.81 0.23 Vox-Fusion [123] 24.41 0.80 0.24 Co-SLAM [124] 30.24 0.94 0.25 GS-SLAM [115] 31.56 0.97 0.094 SplaTAM [116] 34.11 0.97 0.10…”
Vox-Fusion: Dense Tracking and Mapping with Voxel-based Neural Implicit Representation (2022, methodology): “Unlike previous works where keyframes are only inserted based on heuristically chosen metrics [31] or at a fixed interval [40], our explicit voxel structure allows us to determine when to insert key-frames by performing an intersection test.”
DUSt3R: Geometric 3D Vision Made Easy (2023, positive): “To handle large-scale scenes, recent approaches [42, 173, 188, 189] introduce geometry priors to the implicit model, leading to much more detailed reconstructions.”
Co-SLAM: Joint Coordinate and Sparse Parametric Encodings for Neural Real-Time SLAM (2023, methodology): “To reduce the computational overhead and improve scalability, NICESLAM [42] adopts a multi-level feature grid for scene representation.”
Continuous 3D Perception Model with Persistent State (2025, positive): “We evaluate scene-level reconstruction on the 7-scenes [67] and NRGBD [4] datasets using accuracy (Acc), completion (Comp), and normal consistency (NC) metrics, as in prior works [4, 81, 82, 86, 104].”