IE-PMMA:Point Cloud Completion Through Inverse Edge-aware Upsampling and Precise Multi-Modal Feature Alignment
Authors: Ran Jia, Junpeng Xue, Shuai Ma, Wenbo Lu, Kelei Wang
IJCAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We evaluate our method on three typical datasets, and the results demonstrate that our IE-PMMA outperforms the existing state-of-the-art methods quantitatively and visually. |
| Researcher Affiliation | Collaboration | Ran Jia1 , Junpeng Xue1, , Shuai Ma2 , Wenbo Lu1 and Kelei Wang1 1Key Laboratory of Advanced Spatial Mechanism and Intelligent Spacecraft, Ministry of Education, School of Aeronautics and Astronautics, Sichuan University 2Cheng Du Aircraft Industrial (Group) Co., Ltd. |
| Pseudocode | No | The paper describes the methodology in prose and architectural diagrams, but does not include any explicitly labeled pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide any explicit statements about releasing source code for the described methodology, nor does it include links to a code repository. |
| Open Datasets | Yes | We evaluate our method on three typical datasets...PCN[Yuan et al., 2018] dataset is a subset of the Shape Net[Chang et al., 2015] dataset...Shape Net34/55 is proposed by Point Tr[Yu et al., 2021]...We further tested our results on the KITTI[Geiger et al., 2013] dataset. |
| Dataset Splits | No | The paper mentions using specific datasets (PCN, ShapeNet, KITTI) and some of their subsets (e.g., car category for KITTI), but it does not explicitly provide the training, validation, or test split percentages or sample counts for these datasets. It relies on implicit standard splits of these common benchmarks. |
| Hardware Specification | No | The paper refers to a 'detailed complexity and resource cost analyses are shown in Table 6' but Table 6 only lists 'Params' and 'FLOPs' without specifying any hardware (e.g., GPU, CPU models, memory). No other section of the paper details the specific hardware used for experiments. |
| Software Dependencies | No | The paper does not provide specific software dependencies with version numbers (e.g., Python 3.x, PyTorch 1.x, CUDA 11.x). It mentions using 'MLP layers' and 'Point Net++' but not their software implementations or versions. |
| Experiment Setup | Yes | We applied the edge points extracted by EPV during the seed points upsampling process...where ยต is a scaling coefficient set as 0.2...In this paper, we choose 32 32 32 as the resolution to pre-train the EPV model...In the training process, we chose MSE and L1 loss as the loss function to train the probability and coordinates prediction model, respectively. |