Food image recognition with deep convolutional features

In this paper, we report the feature obtained from the Deep Convolutional Neural Network boosts food recognition accuracy greatly by integrating it with conventional hand-crafted image features, Fisher Vectors with HoG and Color patches. In the experiments, we have achieved 72.26% as the top-1 accuracy and 92.00% as the top-5 accuracy for the 100-class food dataset, UEC-FOOD100, which outperforms the best classification accuracy of this dataset reported so far, 59.6%, greatly.

[1]  Edward J. Delp,et al.  Combining global and local features for food identification in dietary assessment , 2011, 2011 18th IEEE International Conference on Image Processing.

[2]  Andrew Zisserman,et al.  Return of the Devil in the Details: Delving Deep into Convolutional Nets , 2014, BMVC.

[3]  Ming Ouhyoung,et al.  Automatic Chinese food identification and quantity estimation , 2012, SIGGRAPH Asia Technical Briefs.

[4]  Keiji Yanai,et al.  FoodCam: A real-time food recognition system on a smartphone , 2015, Multimedia Tools and Applications.

[5]  Trevor Darrell,et al.  DeCAF: A Deep Convolutional Activation Feature for Generic Visual Recognition , 2013, ICML.

[6]  Mei Chen,et al.  Food recognition using statistics of pairwise local features , 2010, 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition.

[7]  Koby Crammer,et al.  Adaptive regularization of weight vectors , 2009, Machine Learning.

[8]  Keiji Yanai,et al.  Recognition of Multiple-Food Images by Detecting Candidate Regions , 2012, 2012 IEEE International Conference on Multimedia and Expo.

[9]  Thomas Mensink,et al.  Improving the Fisher Kernel for Large-Scale Image Classification , 2010, ECCV.

[10]  Andrew Zisserman,et al.  Three things everyone should know to improve object retrieval , 2012, 2012 IEEE Conference on Computer Vision and Pattern Recognition.

[11]  Geoffrey E. Hinton,et al.  ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.