Shapecodes: self-supervised feature learning by lifting views to viewgrids. Jayaraman, D., Gao, R., & Grauman, K. ECCV, 2018. ![Shapecodes: self-supervised feature learning by lifting views to viewgrids [pdf]](https://bibbase.org/img/filetypes/pdf.svg "pdf")
Pdf ![Shapecodes: self-supervised feature learning by lifting views to viewgrids [link]](https://bibbase.org/img/filetypes/link.svg "link")
Paper abstract bibtex We introduce an unsupervised feature learning approach that embeds 3D shape information into a single-view image representation. The main idea is a self-supervised training objective that, given only a single 2D image, requires all unseen views of the object to be predictable from learned features. We implement this idea as an encoder-decoder convolutional neural network. The network maps an input image of an unknown category and unknown viewpoint to a latent space, from which a deconvolutional decoder can best "lift" the image to its complete viewgrid showing the object from all viewing angles. Our class-agnostic training procedure encourages the representation to capture fundamental shape primitives and semantic regularities in a data-driven manner—without manual semantic labels. Our results on two widely-used shape datasets show 1) our approach successfully learns to perform "mental rotation" even for objects unseen during training, and 2) the learned latent space is a powerful representation for object recognition, outperforming several existing unsupervised feature learning methods.
@article{jayaraman2018shapecodes, title= {Shapecodes: self-supervised feature learning by lifting views to viewgrids}, author= {{Jayaraman}, {Dinesh} and Gao, Ruohan and Grauman, Kristen}, journal= {ECCV}, year= {2018},
abstract = {We introduce an unsupervised feature learning approach that embeds 3D shape information into a single-view image representation. The main idea is a self-supervised training objective that, given only a single 2D image, requires all unseen views of the object to be predictable from learned features. We implement this idea as an encoder-decoder convolutional neural network. The network maps an input image of an unknown category and unknown viewpoint to a latent space, from which a deconvolutional decoder can best "lift" the image to its complete viewgrid showing the object from all viewing angles. Our class-agnostic training procedure encourages the representation to capture fundamental shape primitives and semantic regularities in a data-driven manner---without manual semantic labels. Our results on two widely-used shape datasets show 1) our approach successfully learns to perform "mental rotation" even for objects unseen during training, and 2) the learned latent space is a powerful representation for object recognition, outperforming several existing unsupervised feature learning methods.},
url_pdf = {/publication/jayaraman-2018-shapecodes/jayaraman-2018-shapecodes.pdf},
url = {https://arxiv.org/abs/1709.00505},
}
Downloads: 0
{"_id":"Q4twfmxJt99jx5WSZ","bibbaseid":"jayaraman-gao-grauman-shapecodesselfsupervisedfeaturelearningbyliftingviewstoviewgrids-2018","author_short":["Jayaraman, D.","Gao, R.","Grauman, K."],"bibdata":{"bibtype":"article","type":"article","title":"Shapecodes: self-supervised feature learning by lifting views to viewgrids","author":[{"propositions":[],"lastnames":["Jayaraman"],"firstnames":["Dinesh"],"suffixes":[]},{"propositions":[],"lastnames":["Gao"],"firstnames":["Ruohan"],"suffixes":[]},{"propositions":[],"lastnames":["Grauman"],"firstnames":["Kristen"],"suffixes":[]}],"journal":"ECCV","year":"2018","abstract":"We introduce an unsupervised feature learning approach that embeds 3D shape information into a single-view image representation. The main idea is a self-supervised training objective that, given only a single 2D image, requires all unseen views of the object to be predictable from learned features. We implement this idea as an encoder-decoder convolutional neural network. The network maps an input image of an unknown category and unknown viewpoint to a latent space, from which a deconvolutional decoder can best \"lift\" the image to its complete viewgrid showing the object from all viewing angles. Our class-agnostic training procedure encourages the representation to capture fundamental shape primitives and semantic regularities in a data-driven manner—without manual semantic labels. Our results on two widely-used shape datasets show 1) our approach successfully learns to perform \"mental rotation\" even for objects unseen during training, and 2) the learned latent space is a powerful representation for object recognition, outperforming several existing unsupervised feature learning methods.","url_pdf":"/publication/jayaraman-2018-shapecodes/jayaraman-2018-shapecodes.pdf","url":"https://arxiv.org/abs/1709.00505","bibtex":"@article{jayaraman2018shapecodes, title= {Shapecodes: self-supervised feature learning by lifting views to viewgrids}, author= {{Jayaraman}, {Dinesh} and Gao, Ruohan and Grauman, Kristen}, journal= {ECCV}, year= {2018},\n abstract = {We introduce an unsupervised feature learning approach that embeds 3D shape information into a single-view image representation. The main idea is a self-supervised training objective that, given only a single 2D image, requires all unseen views of the object to be predictable from learned features. We implement this idea as an encoder-decoder convolutional neural network. The network maps an input image of an unknown category and unknown viewpoint to a latent space, from which a deconvolutional decoder can best \"lift\" the image to its complete viewgrid showing the object from all viewing angles. Our class-agnostic training procedure encourages the representation to capture fundamental shape primitives and semantic regularities in a data-driven manner---without manual semantic labels. Our results on two widely-used shape datasets show 1) our approach successfully learns to perform \"mental rotation\" even for objects unseen during training, and 2) the learned latent space is a powerful representation for object recognition, outperforming several existing unsupervised feature learning methods.},\n url_pdf = {/publication/jayaraman-2018-shapecodes/jayaraman-2018-shapecodes.pdf},\n url = {https://arxiv.org/abs/1709.00505},\n}\n","author_short":["Jayaraman, D.","Gao, R.","Grauman, K."],"key":"jayaraman2018shapecodes","id":"jayaraman2018shapecodes","bibbaseid":"jayaraman-gao-grauman-shapecodesselfsupervisedfeaturelearningbyliftingviewstoviewgrids-2018","role":"author","urls":{" pdf":"https://gist.githubusercontent.com/publication/jayaraman-2018-shapecodes/jayaraman-2018-shapecodes.pdf","Paper":"https://arxiv.org/abs/1709.00505"},"metadata":{"authorlinks":{}}},"bibtype":"article","biburl":"https://gist.githubusercontent.com/dineshj1/0185709a89b3de5cb7c763e36c0cb031/raw/","dataSources":["EXJChWsnQMWJFyoek","SxhWpwK6aDYTCTgkT","azLig8QfcbeEgFsYA"],"keywords":[],"search_terms":["shapecodes","self","supervised","feature","learning","lifting","views","viewgrids","jayaraman","gao","grauman"],"title":"Shapecodes: self-supervised feature learning by lifting views to viewgrids","year":2018}