AI Paper Atlas

Research map / Point clouds / Point cloud classification

NeuralPCI: Spatio-Temporal Neural Field for 3D Point Cloud Multi-Frame Non-Linear Interpolation

Zehan Zheng, Danni Wu, Ruisi Lu, Fan Lu, Guang Chen, Changjun Jiang

CVPR 2023 · Point clouds › Point cloud classification · approach: Scene flow and dynamic point clouds · 20 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 a multi-frame neural field rather than the strongest pair-frame baselinesNeuralPCI · DHB (Overall)CD: 0.92 -> 0.54Improved
Use a multi-frame neural field rather than the strongest pair-frame baselinesNeuralPCI · DHB (Overall)EMD: 6.14 -> 3.68Improved
Use a multi-frame neural field rather than PointINetNeuralPCI · NL-Drive (Average)CD: 1.06 -> 0.80Improved
Use a multi-frame neural field rather than PointINetNeuralPCI · NL-Drive (Average)EMD: 101.12 -> 97.03Improved
Add multi-frame integration (ablation)NeuralPCI · DHBCD: 0.65 -> 0.59Improved
Add multi-frame integration (ablation)NeuralPCI · NL-DriveEMD: 114.31 -> 112.03Improved
Use the nearest neighbor in time as the reference frame (ablation)NeuralPCI · DHBEMD: 5.08 -> 4.28Improved
Add EMD loss (ablation)NeuralPCI · DHBEMD: 4.28 -> 3.84Improved
Add smoothness loss (ablation)NeuralPCI · NL-DriveEMD: 104.71 -> 99.25Improved
Add position encoding, LeakyReLU, and a wider MLP (ablation)NeuralPCI · NL-DriveCD: 0.82 -> 0.80Improved
Use four input frames rather than two (ablation)NeuralPCI · DHBCD: 0.60 -> 0.54Improved
Use six or eight input frames instead of four (ablation)NeuralPCI · NL-DriveCD: 0.80 -> 0.86; 0.80 -> 0.96Hurt

Later papers that compared with this method

4 later accepted papers report NeuralPCI as a baseline in their results tables; 4 of them report a better number on most of the results they share with it.

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 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