Deep Superpixel Convolutional Network for Image Recognition
被引:8
|
作者:
Zeng, Xianfang
论文数: 0引用数: 0
h-index: 0
机构:
Zhejiang Univ, Inst Cyber Syst & Control, Hangzhou 310027, Peoples R ChinaZhejiang Univ, Inst Cyber Syst & Control, Hangzhou 310027, Peoples R China
Zeng, Xianfang
[1
]
Wu, Wenxuan
论文数: 0引用数: 0
h-index: 0
机构:
Oregon State Univ, CORIS Inst, Corvallis, OR 97331 USAZhejiang Univ, Inst Cyber Syst & Control, Hangzhou 310027, Peoples R China
Wu, Wenxuan
[2
]
Tian, Guangzhong
论文数: 0引用数: 0
h-index: 0
机构:
Zhejiang Univ, Inst Cyber Syst & Control, Hangzhou 310027, Peoples R ChinaZhejiang Univ, Inst Cyber Syst & Control, Hangzhou 310027, Peoples R China
Tian, Guangzhong
[1
]
Li, Fuxin
论文数: 0引用数: 0
h-index: 0
机构:
Oregon State Univ, CORIS Inst, Corvallis, OR 97331 USAZhejiang Univ, Inst Cyber Syst & Control, Hangzhou 310027, Peoples R China
Li, Fuxin
[2
]
Liu, Yong
论文数: 0引用数: 0
h-index: 0
机构:
Zhejiang Univ, State Key Lab Ind Control Technol, Hangzhou 310027, Peoples R ChinaZhejiang Univ, Inst Cyber Syst & Control, Hangzhou 310027, Peoples R China
Liu, Yong
[3
]
机构:
[1] Zhejiang Univ, Inst Cyber Syst & Control, Hangzhou 310027, Peoples R China
[2] Oregon State Univ, CORIS Inst, Corvallis, OR 97331 USA
[3] Zhejiang Univ, State Key Lab Ind Control Technol, Hangzhou 310027, Peoples R China
Due to the high representational efficiency, superpixel largely reduces the number of image primitives for subsequent processing. However, superpixel is scarcely utilized in recent methods since its irregular shape is intractable for standard convolutional layer. In this paper, we propose an end-to-end trainable superpixel convolutional network, named SPNet, to learn high-level representation on image superpixel primitives. We start by treating irregular superpixel lattices as a 2D point cloud, where the low-level features inside one superpixel are aggregated to one feature vector. We replace the standard convolutional layer with the PointConv layer to handle the irregular and unordered point cloud. Besides, we propose grid based downsampling strategies to output uniform 2D sampling result. The resulting network largely utilizes the efficiency of superpixel and provides a novel view for image recognition task. Experiments on image recognition task show promising results compared with prominent image classification methods. The visualization of class activation mapping shows great accuracy at object localization and boundary segmentation.