DeepID3: Face Recognition with Very Deep Neural Networks
Citations
5,308 citations
Cites background or methods from "DeepID3: Face Recognition with Very..."
...However, the final model in [27] is quite complicated involving around 200 CNNs....
[...]
...[24, 25, 26, 27], each of which incrementally but steadily increased the performance on LFW and...
[...]
...A number of new ideas were incorporated over this series of papers, including: using multiple CNNs [25], a Bayesian learning framework [4] to train a metric, multi-task learning over classification and verification [24], different CNN architectures which branch a fully connected layer after each convolution layer [26], and very deep networks inspired by [19, 28] in [27]....
[...]
2,366 citations
1,879 citations
1,346 citations
References
55,235 citations
"DeepID3: Face Recognition with Very..." refers background in this paper
...Compared to the VGG net proposed in previous literature [10, 19], we add additional supervisory signals in a number of fullconnection layers branched out from intermediate layers, which helps to learn better mid-level features and makes optimization of a very deep neural network easier....
[...]
...These two architectures are rebuilt from stacked convolution and inception layers proposed in VGG net [10] and GoogLeNet [16] to make them suitable to face recognition....
[...]
...VGG net stacked multiple convolutional layers together to form complex features....
[...]
...C V ] 3 F eb layers) of VGG net [10] and GoogLeNet [16]....
[...]
...Continuous convolution/inception helps to form features with larger receptive fields and more complex nonlinearity while restricting the number of parameters [10]....
[...]
49,914 citations
40,257 citations
14,799 citations
Additional excerpts
...In the proposed two network architectures, rectified linear non-linearity [9] is used for all except pooling layers, and dropout learning [5] is added on the final feature extraction layer....
[...]
6,899 citations
"DeepID3: Face Recognition with Very..." refers methods in this paper
...In the proposed two network architectures, rectified linear non-linearity [9] is used for all except pooling layers, and dropout learning [5] is added on the final feature extraction layer....
[...]