AI Paper Atlas

Research map / 3D vision / 3D generation

Scalable 3D Captioning with Pretrained Models

Tiange Luo, Chris Rockwell, Honglak Lee, Justin Johnson

NeurIPS 2023 · 3D vision › 3D generation · approach: Text-to-3D scene generation · 21 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
Replace crowdsourced annotation with Cap3DObjaverse captioning · Objaverse, 22k objectsA/B win rate: 37.8% ± 0.5% -> 52.3% ± 0.5%Improved
Replace crowdsourced annotation with Cap3DObjaverse captioning · Objaverse, 22k objectsannotation cost per 1k objects: $87.18 -> $8.35Improved
Replace crowdsourced annotation with Cap3DObjaverse captioning · Objaverse, 22k objectsannotation speed: 1.4k / day -> 65k / dayImproved
Add GPT4 consolidation over multiple rendered views (ablation)BLIP2 captioning · Objaverse, full-dataset random 5k splitViLT image retrieval R@5: 24.7 -> 31.9Improved
Add CLIP filtering before GPT4 consolidation (ablation)BLIP2 + GPT4 captioning · Objaverse, full-dataset random 5k splitCLIPScore: 86.3 -> 86.9Improved
Add CLIP filtering before GPT4 consolidation (ablation)BLIP2 + GPT4 captioning · Objaverse, full-dataset random 5k splitViLT image retrieval R@5: 31.9 -> 31.1Mixed
Use two-stage geometry question answering instead of ordinary captioning (ablation)Cap3D on ABO · ABO, 6.4k objectsA/B win rate against Human: 47.3±1% -> 50.2±1%Improved
Use two-stage geometry question answering instead of ordinary captioning (ablation)Cap3D on ABO · ABO, 6.4k objectsA/B win rate against ordinary Cap3D: 56.0±1% win; 37.4±1% loseImproved
Finetune Point·E on 30k Cap3D captions rather than 30k human captionsPoint·E · Objaverse text-to-3DFID: 34.6 -> 33.7Improved
Increase Point·E finetuning data from 30k to 350k Cap3D captionsPoint·E · Objaverse text-to-3DFID: 33.7 -> 32.8Improved
Finetune Shap·E STF on 350k Cap3D captionsShap·E STF · Objaverse text-to-3DCLIPScore: 80.4 -> 79.1Hurt
Finetune Shap·E STF on 350k Cap3D captionsShap·E STF · Objaverse text-to-3DFID: 37.2 -> 35.5Improved

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