Deformable Part Models with CNN Features

Abstract : In this work we report on progress in integrating deep convo-lutional features with Deformable Part Models (DPMs). We substitute the Histogram-of-Gradient features of DPMs with Convolutional Neural Network (CNN) features, obtained from the top-most, fifth, convolutional layer of Krizhevsky's network [8]. We demonstrate that we thereby obtain a substantial boost in performance (+14.5 mAP) when compared to the baseline HOG-based models. This only partially bridges the gap between DPMs and the currently top-performing R-CNN method of [4], suggesting that more radical changes to DPMs may be needed.
Document type :
Conference papers
Liste complète des métadonnées

Cited literature [12 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-01109290
Contributor : Stavros Tsogkas <>
Submitted on : Monday, January 26, 2015 - 4:50:53 AM
Last modification on : Thursday, February 7, 2019 - 5:29:10 PM
Document(s) archivé(s) le : Monday, April 27, 2015 - 10:11:09 AM

File

Savalle_cnndpm_PnA14.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01109290, version 1

Collections

Citation

Pierre-André Savalle, Stavros Tsogkas, George Papandreou, Iasonas Kokkinos. Deformable Part Models with CNN Features. European Conference on Computer Vision, Parts and Attributes Workshop, Sep 2014, Zurich, Switzerland. ⟨hal-01109290⟩

Share

Metrics

Record views

786

Files downloads

691