3D Recurrent Neural Networks with Context Fusion for Point Cloud Semantic Segmentation[ECCV2018]
论文链接 \qquad 本文主要有两个创新点，首先利用逐点金字塔池化来捕获不同尺度的局部上下文信息，然后利用两个方向上的层级顺序循环神经网络来融合更大范围的上下文信息。算法在室内室外3D点云数据集上均取得了较好的效果。 \qquad 网络首先学习逐点的特征，然后使用逐点的金字塔池化获取不同尺度的上下文信息，将不同尺度的局部信息与点的特征拼接。 \qquad 关于金字塔池化，首先将3D空间按照地平面
相关文章
- 1. Recurrent Slice Networks for 3D Segmentation of Point Clouds[CVPR2018]
- 2. Point Cloud 2019
- 3. Taming Recurrent Neural Networks for Better Summarization
- 4. Recurrent Convolutional Neural Networks for Text Classification
- 5. Scheduled Sampling for Sequence Prediction with Recurrent Neural Networks
- 6. Disconnected Recurrent Neural Networks for Text Categorization
- 7. CHAPTER 9 Sequence Processing with Recurrent Networks
- 8. 019 Recurrent Neural Networks
- 9. [译文] PASS3D: Precise and Accelerated Semantic Segmentation for 3D Point Cloud
- 10. 【译文】 PASS3D: Precise and Accelerated Semantic Segmentation for 3D Point Cloud
- 更多相关文章...
- • RSS
元素 - RSS 教程 - • Swift for 循环 - Swift 教程
- • Spring Cloud 微服务实战(三) - 服务注册与发现
- • Composer 安装与使用
相关标签/搜索
每日一句
-
每一个你不满意的现在，都有一个你没有努力的曾经。
欢迎关注本站公众号,获取更多信息
相关文章
- 1. Recurrent Slice Networks for 3D Segmentation of Point Clouds[CVPR2018]
- 2. Point Cloud 2019
- 3. Taming Recurrent Neural Networks for Better Summarization
- 4. Recurrent Convolutional Neural Networks for Text Classification
- 5. Scheduled Sampling for Sequence Prediction with Recurrent Neural Networks
- 6. Disconnected Recurrent Neural Networks for Text Categorization
- 7. CHAPTER 9 Sequence Processing with Recurrent Networks
- 8. 019 Recurrent Neural Networks
- 9. [译文] PASS3D: Precise and Accelerated Semantic Segmentation for 3D Point Cloud
- 10. 【译文】 PASS3D: Precise and Accelerated Semantic Segmentation for 3D Point Cloud