Neural Network based Clothing Style Analysis via Deep Filter Bank

Neural Network based Clothing Style Analysis via Deep Filter Bank

涂家維(CHIA-WEI TU)

指導教授 ： 江振國

國立中正大學/工學院/資訊工程學系/碩士(2016年)

並列摘要

In this paper, deep fi lter bank is combined by convolution neural net-
work(CNN) and fisher vector(FV) for feature extraction. FV is a method
for feature encoding and it has the good results for texture recognition.
CNN, which is one technology of deep learning, is reaching rise and it is a
hot topic in machine learning sector now. And neural network(NN), as with
CNN, is another one technology of deep learning, is used to our proposed
method. According to the above, we provide a method for clothing style
analysis by using deep filter bank to extract texture and material feature
and some NN models to analyze clothing style.

並列關鍵字

Clothing style analysis ； Neural network(NN) ； Con- volution neural nwtwork(CNN) ； Fisher vector(FV) ； Deep fi lter bank

參考文獻

[12] C. M. Huang, C.-P. Wei, and Y. C. F. Wang. Active learning based clothing image

[19] S. Liu, J. Feng, C. Domokos, H. Xu, J. Huang, Z. Hu, and S. Yan. Fashion parsing

[1] J. Bouvrie. 1 introduction notes on convolutional neural networks.

adaptation for describing people based on ne-grained clothing attributes. In 2015

IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages