Modeling Inter and Intra-Class Relations in the Triplet Loss for Zero-Shot Learning. Le Cacheux, Y., Le Borgne, H., & Crucianu, M. In IEEE/CVF International Conference on Computer Vision (ICCV), Seoul, Korea, 2019.
![Modeling Inter and Intra-Class Relations in the Triplet Loss for Zero-Shot Learning [pdf]](https://bibbase.org/img/filetypes/pdf.svg "pdf")
Pdf Supp ![Modeling Inter and Intra-Class Relations in the Triplet Loss for Zero-Shot Learning [link]](https://bibbase.org/img/filetypes/link.svg "link")
Code abstract bibtex
Pdf Supp Code abstract bibtex Recognizing visual unseen classes, i.e. for which no training data is available, is known as Zero Shot Learning (ZSL). Some of the best performing methods apply the triplet loss to seen classes to learn a mapping between visual representations of images and attribute vectors that constitute class prototypes. They nevertheless make several implicit assumptions that limit their performance on real use cases, particularly with fine-grained datasets comprising a large number of classes. We identify three of these assumptions and put forward corresponding novel contributions to address them. Our approach consists in taking into account both inter-class and intra-class relations, respectively by being more permissive with confusions between similar classes, and by penalizing visual samples which are atypical to their class. The approach is tested on four datasets, including the large-scale ImageNet, and exhibitsperformances significantly above recent methods, even generative methods based on more restrictive hypotheses.
@inproceedings{lecacheux2019iccv,
author = {Le Cacheux, Yannick and Le Borgne, Herv{\'e} and Crucianu, Michel},
title = {Modeling Inter and Intra-Class Relations in the Triplet Loss for Zero-Shot Learning},
booktitle = {IEEE/CVF International Conference on Computer Vision (ICCV)},
year = {2019},
address = {Seoul, Korea},
url_PDF = {https://openaccess.thecvf.com/content_ICCV_2019/papers/Le_Cacheux_Modeling_Inter_and_Intra-Class_Relations_in_the_Triplet_Loss_for_ICCV_2019_paper.pdf},
url_supp = {https://openaccess.thecvf.com/content_ICCV_2019/supplemental/Le_Cacheux_Modeling_Inter_and_ICCV_2019_supplemental.pdf},
url_Code = {https://github.com/yannick-lc/iccv2019-triplet-loss},
abstract = {Recognizing visual unseen classes, i.e. for which no training data is available, is known as Zero Shot Learning (ZSL). Some of the best performing methods apply the triplet loss to seen classes to learn a mapping between visual representations of images and attribute vectors that constitute class prototypes. They nevertheless make several implicit assumptions that limit their performance on real use cases, particularly with fine-grained datasets comprising a large number of classes. We identify three of these assumptions and put forward corresponding novel contributions to address them. Our approach consists in taking into account both inter-class and intra-class relations, respectively by being more permissive with confusions between similar classes, and by penalizing visual samples which are atypical to their class. The approach is tested on four datasets, including the large-scale ImageNet, and exhibitsperformances significantly above recent methods, even generative methods based on more restrictive hypotheses.},
keywords = {zero-shot-learning}
}Downloads: 0
{"_id":"njJZ4dp99smL6LbEN","bibbaseid":"lecacheux-leborgne-crucianu-modelinginterandintraclassrelationsinthetripletlossforzeroshotlearning-2019","author_short":["Le Cacheux, Y.","Le Borgne, H.","Crucianu, M."],"bibdata":{"bibtype":"inproceedings","type":"inproceedings","author":[{"propositions":[],"lastnames":["Le","Cacheux"],"firstnames":["Yannick"],"suffixes":[]},{"propositions":[],"lastnames":["Le","Borgne"],"firstnames":["Hervé"],"suffixes":[]},{"propositions":[],"lastnames":["Crucianu"],"firstnames":["Michel"],"suffixes":[]}],"title":"Modeling Inter and Intra-Class Relations in the Triplet Loss for Zero-Shot Learning","booktitle":"IEEE/CVF International Conference on Computer Vision (ICCV)","year":"2019","address":"Seoul, Korea","url_pdf":"https://openaccess.thecvf.com/content_ICCV_2019/papers/Le_Cacheux_Modeling_Inter_and_Intra-Class_Relations_in_the_Triplet_Loss_for_ICCV_2019_paper.pdf","url_supp":"https://openaccess.thecvf.com/content_ICCV_2019/supplemental/Le_Cacheux_Modeling_Inter_and_ICCV_2019_supplemental.pdf","url_code":"https://github.com/yannick-lc/iccv2019-triplet-loss","abstract":"Recognizing visual unseen classes, i.e. for which no training data is available, is known as Zero Shot Learning (ZSL). Some of the best performing methods apply the triplet loss to seen classes to learn a mapping between visual representations of images and attribute vectors that constitute class prototypes. They nevertheless make several implicit assumptions that limit their performance on real use cases, particularly with fine-grained datasets comprising a large number of classes. We identify three of these assumptions and put forward corresponding novel contributions to address them. Our approach consists in taking into account both inter-class and intra-class relations, respectively by being more permissive with confusions between similar classes, and by penalizing visual samples which are atypical to their class. The approach is tested on four datasets, including the large-scale ImageNet, and exhibitsperformances significantly above recent methods, even generative methods based on more restrictive hypotheses.","keywords":"zero-shot-learning","bibtex":"@inproceedings{lecacheux2019iccv,\n author = {Le Cacheux, Yannick and Le Borgne, Herv{\\'e} and Crucianu, Michel},\n title = {Modeling Inter and Intra-Class Relations in the Triplet Loss for Zero-Shot Learning},\n booktitle = {IEEE/CVF International Conference on Computer Vision (ICCV)},\n year = {2019},\n address = {Seoul, Korea},\n url_PDF = {https://openaccess.thecvf.com/content_ICCV_2019/papers/Le_Cacheux_Modeling_Inter_and_Intra-Class_Relations_in_the_Triplet_Loss_for_ICCV_2019_paper.pdf},\n url_supp = {https://openaccess.thecvf.com/content_ICCV_2019/supplemental/Le_Cacheux_Modeling_Inter_and_ICCV_2019_supplemental.pdf},\n url_Code = {https://github.com/yannick-lc/iccv2019-triplet-loss},\n abstract = {Recognizing visual unseen classes, i.e. for which no training data is available, is known as Zero Shot Learning (ZSL). Some of the best performing methods apply the triplet loss to seen classes to learn a mapping between visual representations of images and attribute vectors that constitute class prototypes. They nevertheless make several implicit assumptions that limit their performance on real use cases, particularly with fine-grained datasets comprising a large number of classes. We identify three of these assumptions and put forward corresponding novel contributions to address them. Our approach consists in taking into account both inter-class and intra-class relations, respectively by being more permissive with confusions between similar classes, and by penalizing visual samples which are atypical to their class. The approach is tested on four datasets, including the large-scale ImageNet, and exhibitsperformances significantly above recent methods, even generative methods based on more restrictive hypotheses.},\n keywords = {zero-shot-learning}\n}\n\n","author_short":["Le Cacheux, Y.","Le Borgne, H.","Crucianu, M."],"key":"lecacheux2019iccv","id":"lecacheux2019iccv","bibbaseid":"lecacheux-leborgne-crucianu-modelinginterandintraclassrelationsinthetripletlossforzeroshotlearning-2019","role":"author","urls":{" pdf":"https://openaccess.thecvf.com/content_ICCV_2019/papers/Le_Cacheux_Modeling_Inter_and_Intra-Class_Relations_in_the_Triplet_Loss_for_ICCV_2019_paper.pdf"," supp":"https://openaccess.thecvf.com/content_ICCV_2019/supplemental/Le_Cacheux_Modeling_Inter_and_ICCV_2019_supplemental.pdf"," code":"https://github.com/yannick-lc/iccv2019-triplet-loss"},"keyword":["zero-shot-learning"],"metadata":{"authorlinks":{}},"html":""},"bibtype":"inproceedings","biburl":"https://hleborgne.github.io/files/hleborgne-publications.bib","dataSources":["sJzmxoNKfHCgQoayi"],"keywords":["zero-shot-learning"],"search_terms":["modeling","inter","intra","class","relations","triplet","loss","zero","shot","learning","le cacheux","le borgne","crucianu"],"title":"Modeling Inter and Intra-Class Relations in the Triplet Loss for Zero-Shot Learning","year":2019}