AI Paper Atlas

Research map / 3D vision / 3D generation

Hilbert Distillation for Cross-Dimensionality Networks

Dian Qin, Haishuai Wang, Zhe Liu, HONGJIA XU, Sheng Zhou, Jiajun Bu

NeurIPS 2022 · 3D vision › 3D generation · approach: 2D-to-3D pretrained representations · 3 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 Hilbert-curve feature distillation instead of the strongest listed baselineResNet-50 student distilled from 3DResNet-50 · ActivityNetTop-1 accuracy: 62.88 ± 0.36 -> 63.55 ± 0.28Improved
Use variable-length Hilbert distillation instead of Hilbert distillationResNet-50 student distilled from 3DResNet-50 · ActivityNetTop-1 accuracy: 63.55 ± 0.28 -> 63.71 ± 0.63Improved
Use Hilbert-curve feature distillation instead of the strongest listed baselineVGG16 student distilled from 3DResNet-50 · ActivityNetTop-1 accuracy: 62.27 ± 0.51 -> 63.46 ± 0.49Improved
Use variable-length Hilbert distillation instead of Hilbert distillationVGG16 student distilled from 3DResNet-50 · ActivityNetTop-1 accuracy: 63.46 ± 0.49 -> 64.02 ± 0.85Improved
Use Hilbert-curve feature distillation instead of the strongest listed baselineResNet-50 student distilled from 3DResNet-50 · Large-COVID-19Top-1 accuracy: 83.69 ± 0.32 -> 85.05 ± 0.58Improved
Use variable-length Hilbert distillation instead of Hilbert distillationResNet-50 student distilled from 3DResNet-50 · Large-COVID-19Top-1 accuracy: 85.05 ± 0.58 -> 85.55 ± 0.72Improved
Use Hilbert-curve feature distillation instead of the strongest listed baselineVGG16 student distilled from 3DResNet-50 · Large-COVID-19Top-1 accuracy: 82.85 ± 0.46 -> 84.63 ± 0.38Improved
Use variable-length Hilbert distillation instead of Hilbert distillationVGG16 student distilled from 3DResNet-50 · Large-COVID-19Top-1 accuracy: 84.63 ± 0.38 -> 85.37 ± 0.86Improved
Use Hilbert-curve feature distillation instead of the strongest listed baselineResNet-50 student distilled from 3DResNet-50 · Kinetics-400Top-1 accuracy: 69.59 ± 1.02 -> 70.28 ± 0.36Improved
Use variable-length Hilbert distillation instead of Hilbert distillationResNet-50 student distilled from 3DResNet-50 · Kinetics-400Top-1 accuracy: 70.28 ± 0.36 -> 70.91 ± 0.85Improved
Use Hilbert-curve feature distillation instead of the strongest listed baselineVGG16 student distilled from 3DResNet-50 · Kinetics-400Top-1 accuracy: 68.48 ± 0.59 -> 69.82 ± 0.42Improved
Use variable-length Hilbert distillation instead of Hilbert distillationVGG16 student distilled from 3DResNet-50 · Kinetics-400Top-1 accuracy: 69.82 ± 0.42 -> 70.40 ± 0.73Improved

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