Multi-view Convolutional Neural Networks for 3D Shape Recognition
Citations
9,457 citations
Cites background from "Multi-view Convolutional Neural Net..."
...Therefore, the model needs to be able to capture local structures from nearby points, and the combinatorial interactions among local structures....
[...]
4,802 citations
2,569 citations
Cites background from "Multi-view Convolutional Neural Net..."
...[20, 15, 10] investigate volumetric and multi-view representation of point cloud for 3D object classification....
[...]
2,565 citations
1,742 citations
Cites background from "Multi-view Convolutional Neural Net..."
...Image-based networks are often multi-view, using a set of 2D images rendered from the point cloud at different viewpoints [35, 4, 18]....
[...]
References
73,978 citations
55,235 citations
"Multi-view Convolutional Neural Net..." refers methods in this paper
...With a deeper network architecture (VGG-VD, a network with 16 weight layers from [34]), we achieve 87....
[...]
49,914 citations
49,639 citations
"Multi-view Convolutional Neural Net..." refers methods in this paper
...Another advantage of using 2D representations is that we can leverage (i) advances in image descriptors in the computer vision community [21, 25] and (ii) massive image databases (such as ImageNet [10]) to pre-train our CNN architectures....
[...]
...Another advantage of using 2D representations is that we can leverage (i) advances in image descriptors [22, 26] and (ii) massive image databases (such as ImageNet [9]) to pre-train our CNN architectures....
[...]
...The network is pre-trained on ImageNet images from 1k categories, and then fine-tuned on all 2D views of the 3D shapes in training set....
[...]
...Here we use the VGG-M network trained on ImageNet without any fine-tuning on either sketches or 3D shapes....
[...]
...Using our CNN baseline trained on ImageNet in turn outperforms Fisher Vectors by a significant margin....
[...]
40,257 citations