On the Performance of GoogLeNet and AlexNet Applied to Sketches

Pedro L. Ballester, Ricardo Matsumura de Araújo

AAAI 2016 pp. 1124-1128

doi:10.1609/AAAI.V30I1.10171 /aaai/2016/ballester2016aaai-performance/

Abstract

This work provides a study on how Convolutional Neural Networks, trained to identify objects primarily in photos, perform when applied to more abstract representations of the same objects. Our main goal is to better understand the generalization abilities of these networks and their learned inner representations. We show that both GoogLeNet and AlexNet networks are largely unable to recognize abstract sketches that are easily recognizable by humans. Moreover, we show that the measured efficacy vary considerably across different classes and we discuss possible reasons for this.

PDF AAAI Semantic Scholar

Cite

Text

Ballester and de Araújo. "On the Performance of GoogLeNet and AlexNet Applied to Sketches." AAAI Conference on Artificial Intelligence, 2016. doi:10.1609/AAAI.V30I1.10171

Markdown

[Ballester and de Araújo. "On the Performance of GoogLeNet and AlexNet Applied to Sketches." AAAI Conference on Artificial Intelligence, 2016.](https://mlanthology.org/aaai/2016/ballester2016aaai-performance/) doi:10.1609/AAAI.V30I1.10171

BibTeX

@inproceedings{ballester2016aaai-performance,
  title     = {{On the Performance of GoogLeNet and AlexNet Applied to Sketches}},
  author    = {Ballester, Pedro L. and de Araújo, Ricardo Matsumura},
  booktitle = {AAAI Conference on Artificial Intelligence},
  year      = {2016},
  pages     = {1124-1128},
  doi       = {10.1609/AAAI.V30I1.10171},
  url       = {https://mlanthology.org/aaai/2016/ballester2016aaai-performance/}
}