Prototypical VoteNet for Few-Shot 3D Point Cloud Object Detection
Authors: Shizhen Zhao, Xiaojuan Qi
NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We conduct extensive experiments to demonstrate the effectiveness of Prototypical Vote Net, and our proposed method shows significant and consistent improvements compared to baselines on two benchmark datasets. Our experimental results on these two benchmark datasets show that the proposed model effectively addresses the few-shot 3D point cloud object detection problem, yielding significant improvement over several competitive baseline approaches. |
| Researcher Affiliation | Academia | Shizhen Zhao, Xiaojuan Qi The University of Hong Kong {zhaosz,xjqi}@eee.hku.hk |
| Pseudocode | No | The paper describes the proposed modules and their functions using text and equations, but does not include any structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | This project will be available at https: //shizhen-zhao.github.io/FS3D_page/. Did you include the code, data, and instructions needed to reproduce the main experimental results (either in the supplemental material or as a URL)? [Yes] |
| Open Datasets | Yes | We construct two new benchmark datasets FS-SUNRGBD and FS-Scan Net. Specifically, FS-SUNRGBD is derived from SUNRGBD [34]... FS-Scan Net is derived from Scan Net [5]. |
| Dataset Splits | No | The paper mentions a 'training set' and 'test' evaluation but does not explicitly provide details about a distinct 'validation' dataset split with percentages or counts. |
| Hardware Specification | No | The paper does not explicitly provide specific hardware details such as GPU models, CPU types, or memory used for running experiments in the main text. |
| Software Dependencies | No | The paper mentions using specific loss functions and backbone networks (e.g., Point Net++), but does not provide specific version numbers for any software dependencies like libraries or frameworks. |
| Experiment Setup | Yes | Therefore, the overall loss for Prototypical Vote Net is given by, Ldet = Lcls + α1Lreg + α2Lobj + α3Lvote, (8) where α1, α2, α3 is the coefficients to balance the loss contributions. Coefficient γ. Table 6 shows the effect of momentum coefficient (γ in Equation (5)). Size of Memory Bank. Table 5 studies the size of the memory bank containing the geometric prototypes. The value of K is set to {30, 60, 90, 120, 150}. |