Frankenstein: Learning Deep Face Representations using Small Data

Abstract : Deep convolutional neural networks have recently proven extremely effective for difficult face recognition problems in uncontrolled settings. To train such networks, very large training sets are needed with millions of labeled images. For some applications, such as near-infrared (NIR) face recognition, such large training datasets are not publicly available and difficult to collect. In this work, we propose a method to generate very large training datasets of synthetic images by compositing real face images in a given dataset. We show that this method enables to learn models from as few as 10,000 training images, which perform on par with models trained from 500,000 images. Using our approach we also obtain state-of-the-art results on the CASIA NIR-VIS2.0 heterogeneous face recognition dataset.
Document type :
Journal articles
Complete list of metadatas

Cited literature [66 references]  Display  Hide  Download


https://hal.inria.fr/hal-01306168
Contributor : Thoth Team <>
Submitted on : Monday, September 25, 2017 - 9:41:55 AM
Last modification on : Wednesday, March 20, 2019 - 1:29:02 AM

Files

bare_jrnl.pdf
Files produced by the author(s)

Identifiers

Collections

Citation

Guosheng Hu, Xiaojiang Peng, Yongxin Yang, Timothy Hospedales, Jakob Verbeek. Frankenstein: Learning Deep Face Representations using Small Data. IEEE Transactions on Image Processing, Institute of Electrical and Electronics Engineers, 2018, 27 (1), pp.293-303. ⟨10.1109/TIP.2017.2756450⟩. ⟨hal-01306168v2⟩

Share

Metrics

Record views

1432

Files downloads

1123