Hi&nbsp;Jérôme,<div><br></div><div>I cloned the repo and tried running test_all.py,</div><div>Seems there are a couple bugs in test_image_functions.py that prevent it from executing properly.</div><div><br></div><div>Is there an example somewhere that I can play with/</div><div><br></div><div>Cheers,</div><div>Marc</div><div><br></div><div><br>On Wednesday, June 12, 2013 10:51:02 PM UTC+2, Jerome Kieffer wrote:<blockquote class="gmail_quote" style="margin: 0;margin-left: 0.8ex;border-left: 1px #ccc solid;padding-left: 1ex;">Dear Pythonistas,
<br>
<br>We are porting the SIFT keypoints extraction algorithm (available from IPOL)
<br>to GPU using PyOpenCL. For the moment, the keypoint location works and
<br>shows a speed-up of 5 to 10x (without tuning so far, vs C++).
<br>
<br>A lot of work is remaining, especially:
<br>* limit the memory footprint (700MB/10Mpix image currently)
<br>* calculate the descriptor for each descriptor
<br>* keypoint matching and image alignment.
<br>* best interleave of IO/CPU/GPU
<br>but we managed to port the most trickiest part to OpenCL (without using
<br>textures, which makes it running also on multi-core).
<br>
<br>I would like to thank the people who published their algorithm on IPOL;
<br>making unit testing possible.
<br>
<br>Last but not least, the code is open source and should have a BSD
<br>licence (even if there is a patent on the algorithm in the USA).
<br><a href="https://github.com/pierrepaleo/sift_pyocl" target="_blank">https://github.com/<wbr>pierrepaleo/sift_pyocl</a>
<br>
<br>Cheers,
<br>
<br>-- 
<br>Jérôme Kieffer &lt;<a href="javascript:" target="_blank" gdf-obfuscated-mailto="3gJArYoWZ40J">goo...@terre-adelie.org</a>&gt;
<br></blockquote></div>