Research map / Point clouds / Shape and surface completion
DiffPCI: Large Motion Point Cloud frame Interpolation with Diffusion Model
Tianyu Zhang, Haobo Jiang, Jian Yang, Jin Xie
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add backward branch (dual branch instead of single) (ablation) | DiffPCI · Nuscenes (average of 3 frames) | Chamfer Distance: 1.12 -> 0.84 | Improved |
| add backward branch (dual branch instead of single) (ablation) | DiffPCI · Argoverse 2 sensor (average of 3 frames) | Chamfer Distance: 0.83 -> 0.67 | Improved |
| add backward branch (dual branch instead of single) (ablation) | DiffPCI · KITTI odometry (average of 3 frames) | Chamfer Distance: 0.58 -> 0.50 | Improved |
| use full-scale encoder instead of DGCNN (ablation) | DiffPCI · Nuscenes (average of 3 frames) | Earth Mover's Distance: 168.55 -> 165.04 | Improved |
| use full-scale encoder instead of DGCNN (ablation) | DiffPCI · Argoverse 2 sensor (average of 3 frames) | Earth Mover's Distance: 64.59 -> 56.79 | Improved |
| use full-scale encoder instead of DGCNN (ablation) | DiffPCI · KITTI odometry (average of 3 frames) | Earth Mover's Distance: 62.17 -> 53.67 | Improved |
| add adaptive selection module with remedial frames (ablation) | DiffPCI · Nuscenes (average of 3 frames) | Earth Mover's Distance: 171.20 -> 165.04 | Improved |
| add adaptive selection module with remedial frames (ablation) | DiffPCI · Argoverse 2 sensor (average of 3 frames) | Earth Mover's Distance: 62.08 -> 56.79 | Improved |
| add adaptive selection module with remedial frames (ablation) | DiffPCI · KITTI odometry (average of 3 frames) | Earth Mover's Distance: 59.51 -> 53.67 | Improved |
| add regularization loss1 (ablation) | DiffPCI · Nuscenes (average of 3 frames) | Earth Mover's Distance: 167.27 -> 165.04 | Improved |
| add regularization loss2 (ablation) | DiffPCI · Argoverse 2 sensor (average of 3 frames) | Earth Mover's Distance: 57.87 -> 56.79 | Improved |
| add regularization loss1 (ablation) | DiffPCI · KITTI odometry (average of 3 frames) | Earth Mover's Distance: 55.35 -> 53.67 | Improved |
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.
- NeuralPCI — NeuralPCI: Spatio-Temporal Neural Field for 3D Point Cloud Multi-Frame Non-Linear Interpolation (CVPR 2023) · better on 6 of 6 shared results
- PointINet — PointINet: Point Cloud Frame Interpolation Network (AAAI 2021) · better on 6 of 6 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
- Hierarchical Flow Diffusion for Efficient Frame Interpolation (CVPR 2025 · similarity 0.88)
- NoiseDiffusion: Correcting Noise for Image Interpolation with Diffusion Models beyond Spherical Linear Interpolation (ICLR 2024 · similarity 0.87)
- SuperPC: A Single Diffusion Model for Point Cloud Completion, Upsampling, Denoising, and Colorization (CVPR 2025 · similarity 0.87)
- High-Resolution Frame Interpolation with Patch-based Cascaded Diffusion (AAAI 2025 · similarity 0.87)
- DiffSF: Diffusion Models for Scene Flow Estimation (NeurIPS 2024 · similarity 0.87)
- EDEN: Enhanced Diffusion for High-quality Large-motion Video Frame Interpolation (CVPR 2025 · similarity 0.87)
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.
