You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "Feynman Liang (JIRA)" <ji...@apache.org> on 2015/06/20 01:12:00 UTC
[jira] [Updated] (SPARK-8491) DAISY Feature Transformer
[ https://issues.apache.org/jira/browse/SPARK-8491?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Feynman Liang updated SPARK-8491:
---------------------------------
Description:
DAISY (Tola et al, PAMI 2010, http://infoscience.epfl.ch/record/138785/files/tola_daisy_pami_1.pdf) is another local image descriptor utilizing histograms of local orientation similar to SIFT. However, one key difference is that the weighted sum of gradient norms used in SIFT's orientation assignment is replaced by convolution with Gaussian kernels. This provides a significant speedup in computing dense descriptors.
We can implement DAISY in Spark ML pipelines as a org.apache.spark.ml.Transformer. Given an image Array[Array[Numeric]], the SIFT transformer should output an ArrayArray[[Numeric]] of the DAISY features for the provided image.
The convolution operation can leverage GPU parallelism for efficiency.
A C++/MATLAB reference implementation is available at http://cvlab.epfl.ch/software/daisy.
was:
DAISY is another local image descriptor utilizing histograms of local orientation similar to SIFT. However, one key difference is that the weighted sum of gradient norms used in SIFT's orientation assignment is replaced by convolution with Gaussian kernels. This provides a significant speedup in computing dense descriptors.
We can implement DAISY in Spark ML pipelines as a org.apache.spark.ml.Transformer. Given an image Array[Array[Numeric]], the SIFT transformer should output an ArrayArray[[Numeric]] of the DAISY features for the provided image.
The convolution operation can leverage GPU parallelism for efficiency.
> DAISY Feature Transformer
> -------------------------
>
> Key: SPARK-8491
> URL: https://issues.apache.org/jira/browse/SPARK-8491
> Project: Spark
> Issue Type: Sub-task
> Components: ML
> Reporter: Feynman Liang
>
> DAISY (Tola et al, PAMI 2010, http://infoscience.epfl.ch/record/138785/files/tola_daisy_pami_1.pdf) is another local image descriptor utilizing histograms of local orientation similar to SIFT. However, one key difference is that the weighted sum of gradient norms used in SIFT's orientation assignment is replaced by convolution with Gaussian kernels. This provides a significant speedup in computing dense descriptors.
> We can implement DAISY in Spark ML pipelines as a org.apache.spark.ml.Transformer. Given an image Array[Array[Numeric]], the SIFT transformer should output an ArrayArray[[Numeric]] of the DAISY features for the provided image.
> The convolution operation can leverage GPU parallelism for efficiency.
> A C++/MATLAB reference implementation is available at http://cvlab.epfl.ch/software/daisy.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org