Francesco Ragusa¹, Valeria Tomaselli², Antonino Furnari¹, Sebastiano Battiato¹, Giovanni Maria Farinella¹

¹University of Catania - Department of Mathematics and Computer Science

²STMicroelectronics - Advanced System Technology - Computer Vision

francescoragusa@outlook.com, valeria.tomaselli@st.com, {furnari,battiato,gfarinella}@dmi.unict.it

Food

Models

In the following, you can download the models which leaded to the best results in our experiments.

• Caffe AlexNet model: network model (prototxt), network weights (caffemodel).
• LibSVM model for binary SVM classification: svm model.

The fine-tuned caffe model could be used directly for classification purposes. However, please note that such model achieves an accuracy of 86,41% in our experiments. Best results are obtained extracting features from the fc7 layer of the provided network and feeding the resulting vector to LibSVM using the provided model.

Please cite the paper below if you use the provided model:

• F. Ragusa, V. Tomaselli, A. Furnari, S. Battiato, G. M. Farinella, “Food vs Non-Food Classification”, 2^nd International Workshop on Multimedia Assisted Dietary Management (MADiMa) in Conjunction with ACM Multimedia, DOI: http://dx.doi.org/10.1145/2986035.2986041, Amsterdam, 16 Oct 2016 (paper, poster)

Related Works