3D Recurrent Neural Networks with Context Fusion for Point Cloud Semantic Segmentation[ECCV2018]
論文鏈接 \qquad 本文主要有兩個創新點，首先利用逐點金字塔池化來捕獲不同尺度的局部上下文信息，然後利用兩個方向上的層級順序循環神經網絡來融合更大範圍的上下文信息。算法在室內室外3D點雲數據集上均取得了較好的效果。 \qquad 網絡首先學習逐點的特徵，然後使用逐點的金字塔池化獲取不同尺度的上下文信息，將不同尺度的局部信息與點的特徵拼接。 \qquad 關於金字塔池化，首先將3D空間按照地平面
相關文章
- 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. [譯文] PASS3D: Precise and Accelerated Semantic Segmentation for 3D Point Cloud
- 10. 【譯文】 PASS3D: Precise and Accelerated Semantic Segmentation for 3D Point Cloud
- 更多相關文章...
- • RSS
元素 - RSS 教程 - • Swift for 循環 - Swift 教程
- • Spring Cloud 微服務實戰(三) - 服務註冊與發現
- • Composer 安裝與使用
相關標籤/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。
歡迎關注本站公眾號,獲取更多信息
相關文章
- 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. [譯文] PASS3D: Precise and Accelerated Semantic Segmentation for 3D Point Cloud
- 10. 【譯文】 PASS3D: Precise and Accelerated Semantic Segmentation for 3D Point Cloud