AI Paper Atlas

Research map / 3D vision / 3D generation

Voxel-based 3D Detection and Reconstruction of Multiple Objects from a Single Image

Feng Liu, Xiaoming Liu

NeurIPS 2021 · 3D vision › 3D generation · approach: Learning 3D reconstruction from images · 15 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Use voxel-aligned 3D features with joint detection and reconstructionProposed framework · ShapeNet-tripletsmean IoU: 43.9 -> 46.0Improved
Use voxel-aligned 3D features with joint detection and reconstructionProposed framework · ShapeNet-pairsglobal IoU: 52.7 -> 55.1Improved
Predict object centers as 3D keypoints rather than 2D image keypoints (ablation)CenterNet-3D · ScanNet-MDRmAP @0.15: 19.5 -> 22.8Improved
Replace DeepLS with local PCA-SDF (ablation)Proposed joint framework · ScanNet-MDRglobal IoU: 35.4 -> 38.2Improved
Add positional encoding to voxel features (ablation)Proposed framework · ScanNet-MDRmAP @0.15: 21.2 -> 22.8Improved
Add positional encoding to voxel features (ablation)Proposed framework · ScanNet-MDRglobal IoU: 37.2 -> 38.2Improved
Jointly train detection and reconstruction instead of training reconstruction alone (ablation)Proposed framework · ScanNet-MDRglobal IoU: 36.5 -> 38.2Improved
Jointly train detection and reconstruction instead of training detection alone (ablation)Proposed framework · ScanNet-MDRmAP @0.15: 20.7 -> 22.8Improved
Reduce voxel size from 0.64m to 0.16m at lB=64 (ablation)Proposed framework · ScanNet-MDRmAP: 18.1 -> 22.8Improved
Reduce voxel size from 0.64m to 0.16m at lB=64 (ablation)Proposed framework · ScanNet-MDRIoU: 31.4 -> 38.2Improved
Increase PCA-SDF latent code size from 32 to 125 (ablation)PCA-SDF · ShapeNet test setmean Chamfer Distance-L2 (×10^3): 0.029 -> 0.022Improved
Replace DeepLS local decoder with PCA-SDF bases (ablation)Shape reconstruction · ShapeNet test setinference time: 0.8081 -> 0.0132Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy