tag:blogger.com,1999:blog-6061887630060661987.post5573165602632050386..comments2024-03-13T13:54:35.259-04:00Comments on dlib C++ Library: Correctly Mirroring DatasetsDavis Kinghttp://www.blogger.com/profile/16577392965630448489noreply@blogger.comBlogger20125tag:blogger.com,1999:blog-6061887630060661987.post-37867664799873163272018-12-11T21:14:53.825-05:002018-12-11T21:14:53.825-05:00You can pass a lambda or any callable object. Put...You can pass a lambda or any callable object. Put whatever state you want in that lambda or callable object.Davis Kinghttps://www.blogger.com/profile/16577392965630448489noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-79707291531322829142018-06-02T03:16:31.790-04:002018-06-02T03:16:31.790-04:00Hello Davis,
At the beginning above all, thank yo...Hello Davis,<br /><br />At the beginning above all, thank you always for your development and great support on your dlib.<br /><br />I am trying to use my face data set with landmark points in the face_landmark_detection_ex.<br />Although I have generated an .xml file of the bounding boxes and landmark positions of faces,<br />I am not sure how to generate a .dat file from that file. <br />I would like to call like:<br />face_landmark_detection_ex 'filename.dat' 'imagename'<br /><br />Is there any way to create .dat file from .xml file with command prompt ?<br /><br />Thank you in advance.Anonymoushttps://www.blogger.com/profile/00353358987342554300noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-31372697311692195372018-05-21T09:47:59.276-04:002018-05-21T09:47:59.276-04:00Run imglab from the same folder as the xml file an...Run imglab from the same folder as the xml file and it will preserve relative paths.Davis Kinghttps://www.blogger.com/profile/16577392965630448489noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-15552656643793820932018-05-21T02:04:06.489-04:002018-05-21T02:04:06.489-04:00Hi Davis,
I have used --flip option on imglab rec...Hi Davis,<br /><br />I have used --flip option on imglab recently.<br />But I found that the image file path in .xml is absolute path instead of relative path.<br />Do you have any suggestions to fix this problem?<br />Thank you!Anonymoushttps://www.blogger.com/profile/04311123924682272090noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-2679971035391495002018-04-20T09:15:55.656-04:002018-04-20T09:15:55.656-04:00Hi Davis, Thanks for your fast reply.
Phew. at le...Hi Davis, Thanks for your fast reply.<br /><br />Phew. at least data are right :) <br />I have found render_face_detections.h which has fixed values as As I understand this one is responsible for rendering the points ? <br /><br />I know that is a silly question but I need to extract the found data and also show that they are found and I'm not that good with c++ at this moment ( mostly java on junior level)<br /><br /><br /><br />Anonymoushttps://www.blogger.com/profile/01952266660173006972noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-71383704292740592512018-04-20T07:08:48.576-04:002018-04-20T07:08:48.576-04:00You don't have to do anything to use a differe...You don't have to do anything to use a different annotation format. Just make a xml file with the annotations and run the trainer on it. The only thing that hard codes 68 is the GUI display code for drawing the lines between points. Don't call that code. Davis Kinghttps://www.blogger.com/profile/16577392965630448489noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-2017784511196921162018-04-20T07:06:47.804-04:002018-04-20T07:06:47.804-04:00Hello Davis,
At the beginning many thanks for you...Hello Davis,<br /><br />At the beginning many thanks for your hard work - dlib is simply awesome.<br /><br />I want to ask you how to adjust face landmark detector (yes yes this one over and over again..) to reflect 194 Helen database when it only shows 68 face landmarks..<br /><br />I tried to follow up your post from September 8, 2014 and make fun with train_shape_predictor_ex.cpp but It's still reflect only 68 points <br />I would be very grateful for any hint how to overcome this problem..<br />Thanks in advance :D<br /><br />Anonymoushttps://www.blogger.com/profile/01952266660173006972noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-45317077342534184222018-02-09T06:03:56.860-05:002018-02-09T06:03:56.860-05:00Use the face landmarks that fall on the mouth corn...Use the face landmarks that fall on the mouth corners? Look at the images in this post. The landmarks tell you where the mouth corners are.Davis Kinghttps://www.blogger.com/profile/16577392965630448489noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-13805531281966693892018-02-09T00:10:47.261-05:002018-02-09T00:10:47.261-05:00Hi, I work with dlib and I have a question. I want...Hi, I work with dlib and I have a question. I want to extract the mouth corners x y coordinates. How I can do it???? Help me, pleaseAnonymoushttps://www.blogger.com/profile/05924672442200183084noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-67253686877846914042018-01-29T15:11:30.951-05:002018-01-29T15:11:30.951-05:00I'm referring to this
I'm not sure if the...I'm referring to <a href="https://github.com/davisking/dlib/blob/master/tools/archive/train_face_5point_model.cpp" rel="nofollow">this</a><br /><br />I'm not sure if there is an actual issue or if I'm maybe missing some constraint. I'll make an issue on github just for easier collaboration and reproduction. Pieterhttps://www.blogger.com/profile/00559912500746484447noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-86663630427439462992018-01-29T10:22:38.575-05:002018-01-29T10:22:38.575-05:00If you think this is something wrong with dlib the...If you think this is something wrong with dlib then submit an issue report on github. Include details.Davis Kinghttps://www.blogger.com/profile/16577392965630448489noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-32313661556864299582018-01-29T10:21:45.022-05:002018-01-29T10:21:45.022-05:00I have no idea what you are talking about. What c...I have no idea what you are talking about. What code?Davis Kinghttps://www.blogger.com/profile/16577392965630448489noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-888450018981190542018-01-29T10:02:38.634-05:002018-01-29T10:02:38.634-05:00Thanks for the quick response. I am running the ex...Thanks for the quick response. I am running the exact same code that you provided for the 5 point example, I only modified the assert on 5 points and added an additional swap for the two mouth points.<br /><br />Just commenting out the flipping gives an estimate of 5 hours for the training. Keeping them in doesn't even get to the estimate in a couple of hours. Pieterhttps://www.blogger.com/profile/00559912500746484447noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-22567677436336068562018-01-29T09:39:56.900-05:002018-01-29T09:39:56.900-05:00Flipping the data should have nothing to do with s...Flipping the data should have nothing to do with speed. You probably just have a bug in your code, are running out of RAM and swapping to disk, or forgot to turn on compiler optimizations.Davis Kinghttps://www.blogger.com/profile/16577392965630448489noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-59278175090585400732018-01-29T09:15:46.539-05:002018-01-29T09:15:46.539-05:00Thanks Davis. We are working on adding three more ...Thanks Davis. We are working on adding three more points to the 5-point landmark data included in dlib: left mouth, right mouth and chin.<br /><br />We've already annotated the data, but the training is exceptionally slow when compared to the 5-point example. <br /><br />The function <i>randomly_generate_split_feature</i> takes anything between one to 5 seconds per test split, practically grinding the training to a halt.<br /><br />When we don't flip or rotate (either does this) then it is on-par with the 5-point example.<br /><br />We did flip the new labels (left mouth and right mouth) accordingly.<br /><br />Do you have any tips on what we can attempt?Pieterhttps://www.blogger.com/profile/00559912500746484447noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-35474755091422045552018-01-26T09:31:31.390-05:002018-01-26T09:31:31.390-05:00Ok so I am going to use loss_mean_squared_multiout...Ok so I am going to use loss_mean_squared_multioutput<fc<68,my_network>>; Is that correct? Also, is there a provision for visualizing the graph of the full network - just to make sure it is correct.Azzyhttps://www.blogger.com/profile/14087639792390904569noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-16222540682328600182018-01-26T09:29:22.775-05:002018-01-26T09:29:22.775-05:00This comment has been removed by the author.Azzyhttps://www.blogger.com/profile/14087639792390904569noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-17692309176271990712018-01-26T09:05:29.306-05:002018-01-26T09:05:29.306-05:00Use a mean squared loss I suppose, but there isn&#...Use a mean squared loss I suppose, but there isn't any tutorial for it. I would find a CVPR paper that discusses this and follow that.Davis Kinghttps://www.blogger.com/profile/16577392965630448489noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-2087343534743515792018-01-26T08:49:49.778-05:002018-01-26T08:49:49.778-05:00Which loss layer would you recommend for training ...Which loss layer would you recommend for training a DNN based landmark detector? My labels are 68 point landmarks with X,Y coordinates. Is there a dlib tutorial covering that?Azzyhttps://www.blogger.com/profile/14087639792390904569noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-13092912194966227722018-01-16T21:41:13.682-05:002018-01-16T21:41:13.682-05:00Thanks Davis for this clarifications
Emad OmarThanks Davis for this clarifications<br />Emad OmarAnonymoushttps://www.blogger.com/profile/09332471575870584825noreply@blogger.com