TY - GEN
T1 - Hierarchical zero-shot classification with convolutional neural network features and semantic attribute learning
AU - Markowitz, Jared
AU - Schmidt, Aurora C.
AU - Burlina, Philippe M.
AU - Wang, I. Jeng
PY - 2017/7/19
Y1 - 2017/7/19
N2 - We examine hierarchical approaches to image classification problems that include categories for which we have no training examples. Building on prior work in hierarchical classification that optimizes the trade-off between depth in a tree and accuracy of placement, we compare the performance of multiple formulations of the problem on both previously seen (non-novel) and previously unseen (novel) classes. We use a subset of 150 object classes from the ImageNet ILSVRC2012 data set, for which we have 218 human-annotated semantic attribute labels and for which we compute deep convolutional features using the OVERFEAT network. We quantitatively evaluate several approaches, using input posteriors derived from distances to SVM classifier boundaries as well as input posteriors based on semantic attribute estimation. We find that the relative performances of the methods differ in non-novel and novel applications and achieve information gains in novel applications through the incorporation of attribute-based posteriors.
AB - We examine hierarchical approaches to image classification problems that include categories for which we have no training examples. Building on prior work in hierarchical classification that optimizes the trade-off between depth in a tree and accuracy of placement, we compare the performance of multiple formulations of the problem on both previously seen (non-novel) and previously unseen (novel) classes. We use a subset of 150 object classes from the ImageNet ILSVRC2012 data set, for which we have 218 human-annotated semantic attribute labels and for which we compute deep convolutional features using the OVERFEAT network. We quantitatively evaluate several approaches, using input posteriors derived from distances to SVM classifier boundaries as well as input posteriors based on semantic attribute estimation. We find that the relative performances of the methods differ in non-novel and novel applications and achieve information gains in novel applications through the incorporation of attribute-based posteriors.
UR - http://www.scopus.com/inward/record.url?scp=85027878269&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85027878269&partnerID=8YFLogxK
U2 - 10.23919/MVA.2017.7986834
DO - 10.23919/MVA.2017.7986834
M3 - Conference contribution
AN - SCOPUS:85027878269
T3 - Proceedings of the 15th IAPR International Conference on Machine Vision Applications, MVA 2017
SP - 194
EP - 197
BT - Proceedings of the 15th IAPR International Conference on Machine Vision Applications, MVA 2017
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 15th IAPR International Conference on Machine Vision Applications, MVA 2017
Y2 - 8 May 2017 through 12 May 2017
ER -