jy9387 / bscp

Shape recognition based on the combination of contour and skeleton (codes for our paper on PRL2016)

BSCP

Shape recognition based on the combination of contour information and skeleton information.

Codes for our PRL2016 paper: "Shape Recognition by Bag of Skeleton-associated Contour Parts". [paper]

Requirements:

1. Dataset: we use Animal Dataset [paper] for evaluating. The dataset contains 2000 images, which are divided into 20 classes. We extend the dataset by adding the corresponding skeleton-associated contour maps. Please [download] it and extract it to the "data/" directory for training and testing.

2. Tools: you might download the pdollar's toolbox [download] and the liblinear [download], and put them into the "include/" directory.

3. Interface: please see exp_animal.m

For more technical details please refer to our paper.

If you find bugs or have questions, please contact me. Glad to hear from you.

---

Mail: [email protected]

Homepage: http://jy9387.github.io/