AI Paper Atlas

Research map / Benchmarks / ScanNet

ScanNet leaderboard

148 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on ScanNet. Best mIoU: 79.8 (PPT (Ours), CVPR 2025).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

ScanNet · mIoU

42 papers, each with its best reported number for its own method; higher is better.

32 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

ScanNet · PSNR

37 papers, each with its best reported number for its own method; higher is better.

27 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

ScanNet · mIoU (val)

35 papers, each with its best reported number for its own method; higher is better.

#mIoU ↑MethodPaperTopic
181.1Utonia (f.t.)Utonia: Toward One Encoder for All Point Clouds
ICML 2026 · Table 3 (indoor)
Point cloud classification
280.7Concerto (full fine-tuning)
PTv3-B
Concerto: Joint 2D-3D Self-Supervised Learning Emerges Spatial Representations
NeurIPS 2025 · Main results table, Full fine-
3D generation
379.1GEM
PTv3-PPT
On Geometry-Enhanced Parameter-Efficient Fine-Tuning for 3D Scene Segmentation
NeurIPS 2025 · Table 2
Point cloud classification
478.9PTv3-Sonata + Ours (fine-tuned)
PTv3-Sonata
How Many Tokens Do 3D Point Cloud Transformer Architectures Really Need?
NeurIPS 2025 · Table 1
Point cloud classification
578.6PTv3 + PPT Sup. (f.t.)
PTv3
Towards Large-scale 3D Representation Learning with Multi-dataset Point Prompt Training
CVPR 2024 · Table 3
Point cloud classification
678PTV3 GS + FastAdapter
PTV3
Mitigating Geometric Degradation in Fast DownSampling via FastAdapter for Point Cloud Segmentation
ICCV 2025 · Table 1
Point cloud classification
777.9CDSegNetAn End-to-End Robust Point Cloud Semantic Segmentation Network with Single-Step Conditional Diffusion Models
CVPR 2025 · Table 1
Segmentation and detection
877.9Flash3DFlash3D: Super-scaling Point Transformers through Joint Hardware-Geometry Locality
CVPR 2025 · Table 1
Point cloud classification
977.8FEAST-MambaFEAST-Mamba: FEAture and SpaTial Aware Mamba Network with Bidirectional Orthogonal Fusion for Cross-Modal Point Cloud Segmentation
AAAI 2025 · Table 2
Segmentation and detection
1076.8RelFlexformer
PTv3
RelFlexformer: Efficient Attention Transformers for Integrable Relative Positional Encodings
NeurIPS 2026 · Table 2
Attention and transformers
25 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

ScanNet · AbsRel

26 papers, each with its best reported number for its own method; lower is better.

16 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

ScanNet · LPIPS

21 papers, each with its best reported number for its own method; lower is better.

11 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

ScanNet · SSIM

20 papers, each with its best reported number for its own method; higher is better.

10 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

ScanNet · F-score

14 papers, each with its best reported number for its own method; higher is better.

#F-score ↑MethodPaperTopic
10.902OursLearning Neural Implicit through Volume Rendering with Attentive Depth Fusion Priors
NeurIPS 2023 · Reconstruction Comparisons on
Depth and stereo
20.834MonoInstance
MonoSDF
MonoInstance: Enhancing Monocular Priors via Multi-view Instance Alignment for Neural Rendering and Reconstruction
CVPR 2025 · Table 1
Depth and stereo
30.799H2O-SDF
NeuS
H2O-SDF: Two-phase Learning for 3D Indoor Reconstruction using Object Surface Fields
ICLR 2024 · Table 3D Reconstruction
NeRF and radiance fields
40.781NC-SDFNC-SDF: Enhancing Indoor Scene Reconstruction Using Neural SDFs with View-Dependent Normal Compensation
CVPR 2024 · Table 1
Face and shape
50.77Ours
MonoSDF
Coordinate Quantized Neural Implicit Representations for Multi-view Reconstruction
ICCV 2023 · Table 2
Face and shape
60.7619Prompt Depth Anything
Depth Anything v2 ViT-L
Prompting Depth Anything for 4K Resolution Accurate Metric Depth Estimation
CVPR 2025 · Table 2
Depth and stereo
70.755HelixSurfHelixSurf: A Robust and Efficient Neural Implicit Surface Learning of Indoor Scenes With Iterative Intertwined Regularization
CVPR 2023 · Table 1
Depth and stereo
80.75Ours (MonoSDF)
MonoSDF
Towards Better Gradient Consistency for Neural Signed Distance Functions via Level Set Alignment
CVPR 2023 · Table: Numerical comparison wi
Shape and surface completion
90.694DG-Recon (c-att)
NeuralRecon
DG-Recon: Depth-Guided Neural 3D Scene Reconstruction
ICCV 2023 · Table 1
3D generation
100.683S3P
VolSDF
Self-Supervised Super-Plane for Neural 3D Reconstruction
CVPR 2023 · Table 3
Depth and stereo
4 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

ScanNet · RMSE

12 papers, each with its best reported number for its own method; lower is better.

2 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

ScanNet · δ1

11 papers, each with its best reported number for its own method; higher is better.

#δ1 ↑MethodPaperTopic
198.7AltNeRFAltNeRF: Learning Robust Neural Radiance Field via Alternating Depth-Pose Optimization
AAAI 2024 · Table 3
NeRF and radiance fields
298.1BetterDepth
Depth Anything
BetterDepth: Plug-and-Play Diffusion Refiner for Zero-Shot Monocular Depth Estimation
NeurIPS 2024 · Table 1
Depth and stereo
397.3Ours (512)Pixel-Perfect Depth with Semantics-Prompted Diffusion Transformers
NeurIPS 2025 · Zero-shot relative depth estim
Autoregressive and latent generation
496.4CH3Depth (96K)CH3Depth: Efficient and Flexible Depth Foundation Model with Flow Matching
CVPR 2025 · Table 1
Video segmentation and tracking
595.7Diff2Flow
Stable Diffusion
Diff2Flow: Training Flow Matching Models via Diffusion Model Alignment
CVPR 2025 · Table 2
Fast diffusion sampling
695.6Marigold + ADDP
Marigold
Aligning Generative Denoising with Discriminative Objectives Unleashes Diffusion for Visual Perception
ICLR 2025 · Table 1
Fast diffusion sampling
795.4DepthFM-ID
SD2.1
DepthFM: Fast Generative Monocular Depth Estimation with Flow Matching
AAAI 2025 · Table 2
Fast diffusion sampling
895.1Marigold (w/ ensemble)
Stable Diffusion v2
Repurposing Diffusion-Based Image Generators for Monocular Depth Estimation
CVPR 2024 · Table 1
Depth and stereo
994MERGE-L
FLUX.1-dev
More Than Generation: Unifying Generation and Depth Estimation via Text-to-Image Diffusion Models
NeurIPS 2025 · Table 1
Text-to-image diffusion
1093.7Ours
DiT-XL
Scaling Properties of Diffusion Models For Perceptual Tasks
CVPR 2025 · Table 3
Fast diffusion sampling
1 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.