AdaSFormer: Adaptive Serialized Transformers for Monocular Semantic Scene Completion from Indoor Environments
Not on the map. Why?
Abstract (arXiv)
Indoor monocular semantic scene completion (MSSC) is notably more challenging than its outdoor counterpart due to complex spatial layouts and severe occlusions. While transformers are well suited for modeling global dependencies, their high memory cost and difficulty in reconstructing fine-grained details have limited their use in indoor MSSC. To address these limitations, we introduce AdaSFormer, a serialized transformer framework tailored for indoor MSSC. Our model features three key designs: (1) an Adaptive Serialized Transformer with learnable shifts that dynamically adjust receptive fields; (2) a Center-Relative Positional Encoding that captures spatial information richness; and (3) a Convolution-Modulated Layer Normalization that bridges heterogeneous representations between convolutional and transformer features. Extensive experiments on NYUv2 and Occ-ScanNet demonstrate that AdaSFormer achieves state-of-the-art performance. The code is publicly available at: https://github.com/alanWXZ/AdaSFormer.
Key facts. Cited by 3 papers in the map (data through 7 Oct 2026). Builds on 88 of its 101 references that are in the map. First posted on arXiv 26 Mar 2026.
Builds on
88 of 101 references.
References · 88
- 2017Attention Is All You Needcited by 59,932
- 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,082
- 2019EfficientNet: Rethinking Model Scaling for Convolutional Neural Networkscited by 6,415
- 2022Scalable Diffusion Models with Transformerscited by 5,931
- 2019DeepSDF: Learning Continuous Signed Distance Functions for Shape Representationcited by 2,682
- 2018Occupancy Networks: Learning 3D Reconstruction in Function Spacecited by 2,037
- 2023Zero-1-to-3: Zero-shot One Image to 3D Objectcited by 1,483
- 2024Depth Anything: Unleashing the Power of Large-Scale Unlabeled Datacited by 1,346
- 2016A Point Set Generation Network for 3D Object Reconstruction from a Single Imagecited by 972
- 2020Convolutional Occupancy Networkscited by 760
- 2020A Survey on Visual Transformercited by 653
- 20163D-R2N2: A Unified Approach for Single and Multi-view 3D Object Reconstructioncited by 587
38 more
- 2023Point Transformer V3: Simpler, Faster, Strongercited by 579
- 2023Tri-Perspective View for Vision-Based 3D Semantic Occupancy Predictioncited by 404
- 2021MonoScene: Monocular 3D Semantic Scene Completioncited by 357
- 2023SurroundOcc: Multi-Camera 3D Occupancy Prediction for Autonomous Drivingcited by 310
- 2023VoxFormer: Sparse Voxel Transformer for Camera-based 3D Semantic Scene Completioncited by 306
- 2023OccFormer: Dual-path Transformer for Vision-based 3D Semantic Occupancy Predictioncited by 287
- 2021Voxel Transformer for 3D Object Detectioncited by 247
- 2023OccWorld: Learning a 3D Occupancy World Model for Autonomous Drivingcited by 230
- 2020Sparse Single Sweep LiDAR Point Cloud Segmentation via Learning Contextual Shape Priors from Scene Completioncited by 222
- 2022Unifying Voxel-based Representation with Transformer for 3D Object Detectioncited by 202
- 2020LMSCNet: Lightweight Multiscale 3D Semantic Completioncited by 187
- 2023CLIP2Scene: Towards Label-efficient 3D Scene Understanding by CLIPcited by 178
- 2024GaussianFormer: Scene as Gaussians for Vision-Based 3D Semantic Occupancy Predictioncited by 155
- 2019xMUDA: Cross-Modal Unsupervised Domain Adaptation for 3D Semantic Segmentationcited by 129
- 2023Robo3D: Towards Robust and Reliable 3D Perception against Corruptionscited by 128
- 2020Anisotropic Convolutional Networks for 3D Semantic Scene Completioncited by 126
- 20203D Sketch-aware Semantic Scene Completion via Semi-supervised Structure Priorcited by 124
- 2023Segment Any Point Cloud Sequences by Distilling Vision Foundation Modelscited by 123
- 2025Are VLMs Ready for Autonomous Driving? An Empirical Study from the Reliability, Data, and Metric Perspectivescited by 120
- 2023OctFormer: Octree-based Transformers for 3D Point Cloudscited by 105
- 2022Voxel Set Transformer: A Set-to-Set Approach to 3D Object Detection from Point Cloudscited by 102
- 2021Holistic 3D Scene Understanding from a Single Image with Implicit Representationcited by 94
- 2022LaserMix for Semi-Supervised LiDAR Semantic Segmentationcited by 92
- 2023Symphonize 3D Semantic Scene Completion with Contextual Instance Queriescited by 88
- 2023NDC-Scene: Boost Monocular 3D Semantic Scene Completion in Normalized Device Coordinates Spacecited by 88
- 2023SCPNet: Semantic Scene Completion on Point Cloudcited by 80
- 2020Semantic Scene Completion using Local Deep Implicit Functions on LiDAR Datacited by 74
- 2023FlatFormer: Flattened Window Attention for Efficient Point Cloud Transformercited by 72
- 2022LargeKernel3D: Scaling up Kernels in 3D Sparse CNNscited by 71
- 2021Panoptic 3D Scene Reconstruction From a Single RGB Imagecited by 70
- 2021ROCA: Robust CAD Model Retrieval and Alignment from a Single Imagecited by 66
- 2024SeeGround: See and Ground for Zero-Shot Open-Vocabulary 3D Visual Groundingcited by 66
- 20253D and 4D World Modeling: A Surveycited by 64
- 2024ComboVerse: Compositional 3D Assets Creation Using Spatially-Aware Diffusion Guidancecited by 59
- 2019RGBD Based Dimensional Decomposition Residual Network for 3D Semantic Scene Completioncited by 54
- 2023Towards Label-free Scene Understanding by Vision Foundation Modelscited by 54
- 2024Multi-Modal Data-Efficient 3D Scene Understanding for Autonomous Drivingcited by 53
- 2023FRNet: Frustum-Range Networks for Scalable LiDAR Segmentationcited by 50
50 most cited of 88 All 88 (JSON)
Built on by
Citing papers per year: 2026: 3.
3 papers.