Mailman 3 June 2013 - scikit-image

GSOC
by Marc de Klerk June 11, 2013

June 11, 2013

Hey guys, The 3 GSOC students, Chintak, Ankit and myself have created a channel, #skimage on freenode. There is an official GSOC channel #gsoc, but this is just to discuss a fews things pertaining to scikit-image in realtime and get to know one another a bit better etc... So please do join in if you can, we promise not to fl00d! Cheers, Marc

3 3

Re: Feature Detectors and Descriptors in scikit-image
by Stéfan van der Walt June 7, 2013

June 7, 2013

Hi Ankit On Fri, Jun 7, 2013 at 3:32 PM, Ankit Agrawal <aaaagrawal(a)gmail.com> wrote: > I took a look at the implementation of Daisy feature descriptor in > skimage and noticed that it finds descriptors around points that are spread > uniformly with density based on step argument as the input. For this I > checked its paper(Pg 3, section 3) and it said it can be used around > feature-points as well as non feature-points. Almost all the Feature > Descriptors(that I know of) including the ones that I am going to implement > are calculated about keypoints. Because of the above reasons, I think an > option should be provided in the functions of feature detectors to return > the output as a vector containing location of feature-points. I would like > to know the views/suggestions of community members experienced in this part > of Computer Vision on this point and to suggest the best possible data-flow > between functions of Feature Detectors and Feature Descriptors. Thank you. I am in favor of developing a feature descriptor API that can calculate descriptors only at specified positions. As far as structures are concerned, we try and stick to ndarrays wherever possible; however, it may not always be possible. Hopefully, we'll discover a good and practical API via your project. Regards Stéfan

1 0

Feature Detectors and Descriptors in scikit-image
by Ankit Agrawal June 7, 2013

June 7, 2013

Hi all, I have some queries about feature Detectors and Descriptors that I have to implement as a part of my GSoC project<https://github.com/scikit-image/scikit-image/wiki/GSoC-2013-Ankit-Agrawal-I…> . Practically, a descriptor is more useful around feature points that are more distinctive in nature and hence will then produce a greater accuracy for the tasks where descriptors are used - Correspondence matching in two image frames for Stereo, Image Alignment, Object Recognition and Tracking etc. On the same note, the feature descriptors in OpenCV take input argument as a Keypoint object which is mainly a vector of keypoints(extracted using a feature detector). OpenCV thus has classes for Keypoint<http://docs.opencv.org/modules/features2d/doc/common_interfaces_of_feature_…>, Feature Detector<http://docs.opencv.org/modules/features2d/doc/common_interfaces_of_feature_…>, DescriptorExtractor<http://docs.opencv.org/modules/features2d/doc/common_interfaces_of_descript…>etc. This enables the flexibility of using any FeatureDescriptor an keypoints extracted using any FeatureDetector. I took a look at the implementation<https://github.com/scikit-image/scikit-image/blob/master/skimage/feature/_d…>of Daisy feature descriptor in skimage and noticed that it finds descriptors around points that are spread uniformly with density based on step argument as the input. For this I checked its paper<http://www.ele.puc-rio.br/~visao/Topicos/Tola%202010%20Efficient%20Dense%20…>(Pg 3, section 3) and it said it can be used around feature-points as well as non feature-points. Almost all the Feature Descriptors(that I know of) including the ones that I am going to implement are calculated about keypoints. Because of the above reasons, I think an option should be provided in the functions of feature detectors to return the output as a vector containing location of feature-points. I would like to know the views/suggestions of community members experienced in this part of Computer Vision on this point and to suggest the best possible data-flow between functions of Feature Detectors and Feature Descriptors. Thank you. Regards, Ankit Agrawal, Communication and Signal Processing, IIT Bombay.

1 0

Re: Help needed with Git
by Ankit Agrawal June 6, 2013

June 6, 2013

Hi Puneeth, Thanks a lot!! Your described workflow worked smoothly and I enjoyed resolving the conflicts. @Others : Please review the PR<https://github.com/scikit-image/scikit-image/pull/505> and comment if any other changes/additions are needed. On Thursday, June 6, 2013 8:45:53 AM UTC+5:30, punchagan wrote: > > Hi Ankit, > > On Thu, Jun 6, 2013 at 8:11 AM, Ankit Agrawal <aaaag...(a)gmail.com<javascript:>> > wrote: > > Hi everyone, > > > > For this PR, I was trying to rebase but messed up my local > > repository because of my noobness with git. My complete log can be > viewed > > here. Probably Stefan got busy over the past two days and hence could > not > > reply to my last query in the discussions of the above PR. It would be > great > > if someone here could guide me resolving my repo. If a chat medium seems > > convenient for you to guide me through this mess, I am available on IRC > > channel #skimage. Thank you. > > Looks like you didn't go through, until the end of the rebase process, > to complete it and that messed up things quite a bit. > > Assuming, you haven't made any more changes, apart from what your bash > session log shows, you should be able to restore world-order by > following these steps: > > 1. Abort the previous rebase > `git rebase --abort` > > 2. If you had any local changes, stash them away > `git stash save messed-up rebase` # everything following save is > just a message > > 3. `git checkout gammaCorrect` > > 4. `git rebase master` > > 5. Now, don't add all the changes, without resolving conflicts! > Look at [1] and [2] for more information on how to resolve conflicts > > 6. Add all the files in which you have resolved conflicts. > `skimage/exposure/__init__.py` seems to be one of them > `git add skimage/exposure/__init__.py` > # Add other files, if any > > 7. `git rebase --continue` > > 8. There might be more conflicts, which will take you back to step 5. > Continue doing 5, 6, 7, until you are done. > > 9. Force push your re-based branch > `git push -f github gammaCorrect` > > 10. You are done! Notify the devs! > > Hope that helps, > Puneeth > > [1] - > http://git-scm.com/book/en/Git-Branching-Basic-Branching-and-Merging#Basic-… > [2] - http://githowto.com/resolving_conflicts >

1 0

Help needed with Git
by Ankit Agrawal June 6, 2013

June 6, 2013

Hi everyone, For this PR <https://github.com/scikit-image/scikit-image/pull/505>, I was trying to rebase but messed up my local repository because of my noobness with git. My complete log can be viewed here<http://pastebin.com/yTpe78M6>. Probably Stefan got busy over the past two days and hence could not reply to my last query in the discussions of the above PR. It would be great if someone here could guide me resolving my repo. If a chat medium seems convenient for you to guide me through this mess, I am available on IRC channel #skimage. Thank you. Regards, Ankit Agrawal, Communication and Signal Processing, IIT Bombay.

2 1

Re: Color combine function
by Ronnie Ghose June 6, 2013

June 6, 2013

Np dstack/hstack/vstack wasn't that bad for me but some of the things you can do with dimensions in numpy and how you do them efficiently / sometimes even do them at all ..wow o_o those still surprise me, .e.g. some of the ways to do things in _geometic.py :) On Wed, Jun 5, 2013 at 8:35 PM, Juan Nunez-Iglesias <jni.soma(a)gmail.com>wrote: > I too was unaware of np.dstack for a long time, and was doing it via > np.concatenate([R[..., np.newaxis], G[..., np.newaxis], B[..., > np.newaxis]], axis=-1)... Which is a mouthful. ;) And even after I became > aware of it I still couldn't remember to use it! Sometimes aliases are > useful. And, as we mentioned, channel handling could be important, though I > usually have three channels. > > > On Thu, Jun 6, 2013 at 4:26 AM, Ronnie Ghose <ronnie.ghose(a)gmail.com>wrote: > >> Its just an alias on top of for ease of use I assume >> On Jun 5, 2013 12:54 PM, "Johannes Schönberger" <jschoenberger(a)demuc.de> >> wrote: >> >>> > I didn't know that one, my bad... So the only thing left is the data >>> handling (float conversion and possibly normalization), plus the cases >>> where you only provide 1 or 2 channels (which often the case in >>> fluorescence images), which is probably not worth a function. Thank's for >>> the tip anyway >>> >>> If the channel handling (when one of the channels is missing) is >>> regarded as a common use case, we could definitely add this function! >>> >>> Johannes Schönberger >>> >>> Am 05.06.2013 um 18:17 schrieb Guillaume Gay < >>> guillaume(a)mitotic-machine.org>: >>> >>> > >>> > Le 05/06/2013 17:34, Johannes Schönberger a écrit : >>> >> np.dstack([R, G, B]) >>> > I didn't know that one, my bad... So the only thing left is the data >>> handling (float conversion and possibly normalization), plus the cases >>> where you only provide 1 or 2 channels (which often the case in >>> fluorescence images), which is probably not worth a function. Thank's for >>> the tip anyway >>> > >>> > G. >>> > >>> > -- >>> > You received this message because you are subscribed to the Google >>> Groups "scikit-image" group. >>> > To unsubscribe from this group and stop receiving emails from it, send >>> an email to scikit-image+unsubscribe(a)googlegroups.com. >>> > For more options, visit https://groups.google.com/groups/opt_out. >>> > >>> > >>> >>> -- >>> You received this message because you are subscribed to the Google >>> Groups "scikit-image" group. >>> To unsubscribe from this group and stop receiving emails from it, send >>> an email to scikit-image+unsubscribe(a)googlegroups.com. >>> For more options, visit https://groups.google.com/groups/opt_out. >>> >>> >>> -- >> You received this message because you are subscribed to the Google Groups >> "scikit-image" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to scikit-image+unsubscribe(a)googlegroups.com. >> For more options, visit https://groups.google.com/groups/opt_out. >> >> >> > > -- > You received this message because you are subscribed to the Google Groups > "scikit-image" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to scikit-image+unsubscribe(a)googlegroups.com. > For more options, visit https://groups.google.com/groups/opt_out. > > >

1 0

Re: Color combine function
by Juan Nunez-Iglesias June 6, 2013

June 6, 2013

I too was unaware of np.dstack for a long time, and was doing it via np.concatenate([R[..., np.newaxis], G[..., np.newaxis], B[..., np.newaxis]], axis=-1)... Which is a mouthful. ;) And even after I became aware of it I still couldn't remember to use it! Sometimes aliases are useful. And, as we mentioned, channel handling could be important, though I usually have three channels. On Thu, Jun 6, 2013 at 4:26 AM, Ronnie Ghose <ronnie.ghose(a)gmail.com> wrote: > Its just an alias on top of for ease of use I assume > On Jun 5, 2013 12:54 PM, "Johannes Schönberger" <jschoenberger(a)demuc.de> > wrote: > >> > I didn't know that one, my bad... So the only thing left is the data >> handling (float conversion and possibly normalization), plus the cases >> where you only provide 1 or 2 channels (which often the case in >> fluorescence images), which is probably not worth a function. Thank's for >> the tip anyway >> >> If the channel handling (when one of the channels is missing) is regarded >> as a common use case, we could definitely add this function! >> >> Johannes Schönberger >> >> Am 05.06.2013 um 18:17 schrieb Guillaume Gay < >> guillaume(a)mitotic-machine.org>: >> >> > >> > Le 05/06/2013 17:34, Johannes Schönberger a écrit : >> >> np.dstack([R, G, B]) >> > I didn't know that one, my bad... So the only thing left is the data >> handling (float conversion and possibly normalization), plus the cases >> where you only provide 1 or 2 channels (which often the case in >> fluorescence images), which is probably not worth a function. Thank's for >> the tip anyway >> > >> > G. >> > >> > -- >> > You received this message because you are subscribed to the Google >> Groups "scikit-image" group. >> > To unsubscribe from this group and stop receiving emails from it, send >> an email to scikit-image+unsubscribe(a)googlegroups.com. >> > For more options, visit https://groups.google.com/groups/opt_out. >> > >> > >> >> -- >> You received this message because you are subscribed to the Google Groups >> "scikit-image" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to scikit-image+unsubscribe(a)googlegroups.com. >> For more options, visit https://groups.google.com/groups/opt_out. >> >> >> -- > You received this message because you are subscribed to the Google Groups > "scikit-image" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to scikit-image+unsubscribe(a)googlegroups.com. > For more options, visit https://groups.google.com/groups/opt_out. > > >

1 0

Re: Color combine function
by Johannes Schönberger June 5, 2013

June 5, 2013

Maybe I misunderstand the purpose of the function but what's wrong with np.dstack([R, G, B])? Johannes Schönberger Am 05.06.2013 um 16:52 schrieb Juan Nunez-Iglesias <jni.soma(a)gmail.com>: > I'd certainly use such a function. I've done it by hand quite a few times. I agree that a two channel option where the user can specify the channels would be useful. I would say, for example, > > def combine_to_rgb(images, channels=(0, 1, 2)): > > do_something() > > > Then the user can specify which channel corresponds to which image. So, for two images, channels=(2, 0), the function would make the first image the blue channel and the second image the red channel. > > > On Thu, Jun 6, 2013 at 12:04 AM, Ronnie Ghose <ronnie.ghose(a)gmail.com> wrote: > just wondering what do you do if there's 2 grey level images? an option to choose which channels? > > > Thanks, > Ronnie > > > On Wed, Jun 5, 2013 at 9:58 AM, Guillaume Gay <guillaume(a)mitotic-machine.org> wrote: > Hi list, > > I coded a very simple color combine function, that creates an RGB image from one to three grey level images. Is there an interest to add it to skimage? Would colorconv.py be the correct place to add it? > > Cheers, > > Guillaume > > -- > You received this message because you are subscribed to the Google Groups "scikit-image" group. > To unsubscribe from this group and stop receiving emails from it, send an email to scikit-image+unsubscribe(a)googlegroups.com. > For more options, visit https://groups.google.com/groups/opt_out. > > > > > -- > You received this message because you are subscribed to the Google Groups "scikit-image" group. > To unsubscribe from this group and stop receiving emails from it, send an email to scikit-image+unsubscribe(a)googlegroups.com. > For more options, visit https://groups.google.com/groups/opt_out. > > > > > -- > You received this message because you are subscribed to the Google Groups "scikit-image" group. > To unsubscribe from this group and stop receiving emails from it, send an email to scikit-image+unsubscribe(a)googlegroups.com. > For more options, visit https://groups.google.com/groups/opt_out. > >

3 3

Re: Color combine function
by Juan Nunez-Iglesias June 5, 2013

June 5, 2013

Haha or Guillaume's way works also, and well. ;) On Thu, Jun 6, 2013 at 12:52 AM, Juan Nunez-Iglesias <jni.soma(a)gmail.com>wrote: > I'd certainly use such a function. I've done it by hand quite a few times. > I agree that a two channel option where the user can specify the channels > would be useful. I would say, for example, > > def combine_to_rgb(images, channels=(0, 1, 2)): > do_something() > > > Then the user can specify which channel corresponds to which image. So, > for two images, channels=(2, 0), the function would make the first image > the blue channel and the second image the red channel. > > > On Thu, Jun 6, 2013 at 12:04 AM, Ronnie Ghose <ronnie.ghose(a)gmail.com>wrote: > >> just wondering what do you do if there's 2 grey level images? an option >> to choose which channels? >> >> >> Thanks, >> Ronnie >> >> >> On Wed, Jun 5, 2013 at 9:58 AM, Guillaume Gay < >> guillaume(a)mitotic-machine.org> wrote: >> >>> Hi list, >>> >>> I coded a very simple color combine function, that creates an RGB image >>> from one to three grey level images. Is there an interest to add it to >>> skimage? Would colorconv.py be the correct place to add it? >>> >>> Cheers, >>> >>> Guillaume >>> >>> -- >>> You received this message because you are subscribed to the Google >>> Groups "scikit-image" group. >>> To unsubscribe from this group and stop receiving emails from it, send >>> an email to scikit-image+unsubscribe(a)**googlegroups.com<scikit-image%2Bunsubscribe(a)googlegroups.com> >>> . >>> For more options, visit https://groups.google.com/**groups/opt_out<https://groups.google.com/groups/opt_out> >>> . >>> >>> >>> >> -- >> You received this message because you are subscribed to the Google Groups >> "scikit-image" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to scikit-image+unsubscribe(a)googlegroups.com. >> For more options, visit https://groups.google.com/groups/opt_out. >> >> >> > >

1 0

Re: Color combine function
by Juan Nunez-Iglesias June 5, 2013

June 5, 2013

I'd certainly use such a function. I've done it by hand quite a few times. I agree that a two channel option where the user can specify the channels would be useful. I would say, for example, def combine_to_rgb(images, channels=(0, 1, 2)): do_something() Then the user can specify which channel corresponds to which image. So, for two images, channels=(2, 0), the function would make the first image the blue channel and the second image the red channel. On Thu, Jun 6, 2013 at 12:04 AM, Ronnie Ghose <ronnie.ghose(a)gmail.com>wrote: > just wondering what do you do if there's 2 grey level images? an option to > choose which channels? > > > Thanks, > Ronnie > > > On Wed, Jun 5, 2013 at 9:58 AM, Guillaume Gay < > guillaume(a)mitotic-machine.org> wrote: > >> Hi list, >> >> I coded a very simple color combine function, that creates an RGB image >> from one to three grey level images. Is there an interest to add it to >> skimage? Would colorconv.py be the correct place to add it? >> >> Cheers, >> >> Guillaume >> >> -- >> You received this message because you are subscribed to the Google Groups >> "scikit-image" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to scikit-image+unsubscribe(a)**googlegroups.com<scikit-image%2Bunsubscribe(a)googlegroups.com> >> . >> For more options, visit https://groups.google.com/**groups/opt_out<https://groups.google.com/groups/opt_out> >> . >> >> >> > -- > You received this message because you are subscribed to the Google Groups > "scikit-image" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to scikit-image+unsubscribe(a)googlegroups.com. > For more options, visit https://groups.google.com/groups/opt_out. > > >

1 0