Boundary-Aware Geometric Encoding for Semantic Segmentation of Point Clouds
Authors: Jingyu Gong, Jiachen Xu, Xin Tan, Jie Zhou, Yanyun Qu, Yuan Xie, Lizhuang Ma1424-1432
AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The experiments can be divided into two parts. We demonstrate the performance of our method and compare it with other state-of-the-art methods on Scan Net v2 (Dai et al. 2017) and S3DIS Area-5 (Armeni et al. 2016) for scene semantic segmentation task, respectively. Then, intensive ablation studies are conducted. We take the mean intersection-over-union (m Io U) over categories as our metric like many previous works (Wu, Qi, and Fuxin 2019). Code is available at https://github.com/Jchen Xu/Boundary Aware GEM. |
| Researcher Affiliation | Academia | 1 Department of Computer Science and Engineering, Shanghai Jiao Tong University, Shanghai, China 2 School of Computer Science and Technology, East China Normal University, Shanghai, China 3 City University of Hong Kong, HKSAR, China 4 School of Informatics, Xiamen University, Fujian, China |
| Pseudocode | No | The paper does not contain structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | Code is available at https://github.com/Jchen Xu/Boundary Aware GEM. |
| Open Datasets | Yes | In scene semantic segmentation task, we evaluate our method on Scan Net v2 (Dai et al. 2017) and S3DIS (Armeni et al. 2016). |
| Dataset Splits | Yes | In Scan Net v2, there are totally 1, 201 scanned scenes for training and 312 scenes for validation. Additionally, another 100 scenes are provided as the testing samples, and there are 20 different categories. Following (Wu, Qi, and Fuxin 2019), we randomly sample 3m 1.5m 1.5m cubes from rooms with 8,192 points as the training samples, and test over the entire scan. In S3DIS, there are six indoor areas including 271 rooms from three different buildings. Each point is annotated with a corresponding label from 13 categories. We split points by room and sample all rooms into 0.5m 0.5m blocks with 0.25m padding. Like experiment setting used in previous works (Qi et al. 2017a; Li et al. 2018), we split Area 5 as the test set and use others for training. In the training areas, 4,096 points are sampled for each block and all points in the testing areas are used for testing block-wisely. |
| Hardware Specification | Yes | Our model is trained by Adam optimizer with batch size 8 for Scan Net and batch size 12 for S3DIS on a GTX 1080Ti GPU. |
| Software Dependencies | No | The paper mentions the Adam optimizer but does not specify software dependencies with version numbers. |
| Experiment Setup | Yes | Our model is trained by Adam optimizer with batch size 8 for Scan Net and batch size 12 for S3DIS on a GTX 1080Ti GPU. Also, we analyze the number of the ground truth of boundary and non-boundary points in different scenes. Accordingly, for Scan Net, w1 and w2 used in LBP M are 1 and 10, and for S3DIS, w1 and w2 are 1 and 2. |