tag:blogger.com,1999:blog-6061887630060661987.post8523159453310683841..comments2024-03-13T13:54:35.259-04:00Comments on dlib C++ Library: Python Stuff and Real-Time Video Object TrackingDavis Kinghttp://www.blogger.com/profile/16577392965630448489noreply@blogger.comBlogger88125tag:blogger.com,1999:blog-6061887630060661987.post-51201223616220277552019-03-26T10:48:18.372-04:002019-03-26T10:48:18.372-04:00Many thanks for confirming that negative coordinat...Many thanks for confirming that negative coordinates are expected. As I drive, the speed signs in the captured video goes out of the image to the top and left. I now handled this exception in my code.Jayhttps://www.blogger.com/profile/05252584215108621326noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-76447896088442210382019-03-24T18:00:12.937-04:002019-03-24T18:00:12.937-04:00I assume it's going out of the image to the le...I assume it's going out of the image to the left or top? Those areas have negative coordinates, so this is expected.Davis Kinghttps://www.blogger.com/profile/16577392965630448489noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-31666019804387496132019-03-24T10:57:39.694-04:002019-03-24T10:57:39.694-04:00Hi Davis,
I am currently using Correlation...Hi Davis,<br /> I am currently using Correlation tracker to track speed limit signs in the videos. The tracker works fine, however when the speed sign goes out of the image the tracker returns negative x and y values. I used <br /><br />tracker.update(current_image)<br />cout<<tracker.get_position().left()<< " "<<tracker.get_position().top()<<" "<<tracker.get_position().right()<<" "<<tracker.get_position().bottom()<<endl;<br /><br />I tested the tracker with three videos and I observe this behaviour whenever the speed sign goes out of the image. Please suggest whether this is an expected behaviour or not. Thanks.Jayhttps://www.blogger.com/profile/05252584215108621326noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-44607900366610479012019-01-10T19:18:25.804-05:002019-01-10T19:18:25.804-05:00There isn't any GPU accelerated version of thi...There isn't any GPU accelerated version of this.Davis Kinghttps://www.blogger.com/profile/16577392965630448489noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-76097903778453864202019-01-10T15:05:27.273-05:002019-01-10T15:05:27.273-05:00Hi David, is possible to use a GPU with this algor...Hi David, is possible to use a GPU with this algorithm ? for one up to 5 objects it's work fine. But add one more and all is going very slow.<br />Best regards.<br />Martín.Anonymoushttps://www.blogger.com/profile/17375719660023102670noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-28733350721876310022019-01-10T15:02:53.134-05:002019-01-10T15:02:53.134-05:00Hi David, is possible to use a GPU with this algor...Hi David, is possible to use a GPU with this algorithm ? for one up to 5 objects it's work fine. But add one more and all is going very slow.<br />Best regards.<br />Martín.Anonymoushttps://www.blogger.com/profile/17375719660023102670noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-65697045715509334472018-02-26T06:54:16.418-05:002018-02-26T06:54:16.418-05:00The video doesn't use any additional processin...The video doesn't use any additional processing tricks. But in general this kind of algorithm will often, but not always, get confused if two similar looking objects briefly occlude each other. To make it more robust to this kind of thing you need to add some stronger appearance based features like pull out a face descriptor and use that to deal with track swaps. There is also an extended version of this algorithm that is better at disambiguating this kind of issue (http://openaccess.thecvf.com/content_cvpr_2017/papers/Mueller_Context-Aware_Correlation_Filter_CVPR_2017_paper.pdf) which was presented at last year's CVPR. I haven't added it to dlib yet though.Davis Kinghttps://www.blogger.com/profile/16577392965630448489noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-617798335490757062018-02-26T04:09:11.276-05:002018-02-26T04:09:11.276-05:00Hi Davis,
I am particular interested in the part ...Hi Davis,<br /><br />I am particular interested in the part as shown in the video from 3:32 - 3:44, where two person cross and intersect. <br /><br />I have implement the tracker where I track faces, the tracker works very well with one face in the screen. When I tried to track two face where they crosses, when the tracked face is in the front it is still working good. However when the tracked face is in the back, the front face will 'bring' the tracker away and the tracked object will now become the face in the back.<br /><br />I am running the tracker at around a 30fps video (real time from webcam). Unlike shown in the video where the tracker will still recognize the tracking object even when two person cross and intersects, regardless the tracking object is in the back or the front. <br /><br />Is there any additional algorithm applied in order to achieve the performance as shown in the video? As my understanding is that the tracker algorithm will look at the closest pixel in the bounding box of current and subsequence frame, hence what I observed from my implementation should be correct.<br /><br />Thank you.<br /><br />Regards<br />YZYZ Rhttps://www.blogger.com/profile/09452029903981103668noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-70738343225489405992017-08-11T05:07:02.028-04:002017-08-11T05:07:02.028-04:00That is a good tracker. However, it is unfortunate...That is a good tracker. However, it is unfortunately not possible to track multiple objects.<br /><br />In my case there is many tiny objects which I do not have any problems detecting them. However, when it comes to tracking. I have no clues how to do so.<br /><br />how exactly one is able to create multiple tracker.update() while having not much speed performance degradation?<br />I mean how is it possible to use parallel processing in GPU track each object individually using parallel processing.<br />I have tried to reproduce such a system: https://www.youtube.com/watch?v=3IR1h6i31JI<br />so far detection is good, but tracking fails miserably.Anonymoushttps://www.blogger.com/profile/13060767218066286333noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-37844360009251827352017-08-10T15:30:15.064-04:002017-08-10T15:30:15.064-04:00This comment has been removed by the author.Anonymoushttps://www.blogger.com/profile/13060767218066286333noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-8140664647277566172017-08-10T15:19:51.851-04:002017-08-10T15:19:51.851-04:00This comment has been removed by the author.Anonymoushttps://www.blogger.com/profile/13060767218066286333noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-74498826591065325432017-05-09T14:01:28.185-04:002017-05-09T14:01:28.185-04:00hi davis, i tried to use dlib to place face andmar...hi davis, i tried to use dlib to place face andmarks but it gives out this error henever i try to use shape_pridictor:<br /><br />error : predictor = dlib.shape_predictor(predictor_path)<br />RuntimeError: Error deserializing a floating point number.<br />while deserializing a dlib::matrix<br />while deserializing object of type std::vector<br />while deserializing object of type std::vector<br />while deserializing object of type std::vector<br /><br />i am using latest version on my raspberry pi.<br />i need to get this done because i have a presentation for my final year project please i need help !Celebnews Cornerhttps://www.blogger.com/profile/14551569141423763544noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-92108177882343794302017-05-09T13:57:55.306-04:002017-05-09T13:57:55.306-04:00This comment has been removed by the author.Celebnews Cornerhttps://www.blogger.com/profile/14551569141423763544noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-21234680123004948712017-01-30T22:23:43.822-05:002017-01-30T22:23:43.822-05:00Hi Davis
Thanks for creating Dlib. I found it rea...Hi Davis<br /><br />Thanks for creating Dlib. I found it really useful.<br /><br />I have a couple questions about correation_trakcer:<br /><br />1) How can I obtain relevant values of tracking-rectangle (e.g., center position, width, height, etc.)? It seems that I need to do something with get_position() but I cannot get the value I want.<br /><br />2) When an tracked-object moves fast, the tracker can lose the object. Is there any way to minimize this possibility? (I know that I need to detect the object again in case it is lost.)<br /><br />Thanks in advance.Anonymoushttps://www.blogger.com/profile/03851281540861900486noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-26417925558958011772016-07-30T07:03:17.828-04:002016-07-30T07:03:17.828-04:00Really cool!Really cool!Anonymoushttps://www.blogger.com/profile/17579375577821592271noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-90497761377367187372016-07-12T13:29:20.035-04:002016-07-12T13:29:20.035-04:00You have to use C++ to do that.You have to use C++ to do that.Davis Kinghttps://www.blogger.com/profile/16577392965630448489noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-72364095392984462672016-07-12T13:04:24.642-04:002016-07-12T13:04:24.642-04:00Great work. Can I modify the tracking parameters i...Great work. Can I modify the tracking parameters in python or do I need to recompile dlib for each change? dapper danhttps://www.blogger.com/profile/11099999085375813861noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-83248694544510730142016-05-14T06:25:21.713-04:002016-05-14T06:25:21.713-04:00Those are good ideas and you will need to test the...Those are good ideas and you will need to test them out when you develop your system to see what works. That is the only way to know.Davis Kinghttps://www.blogger.com/profile/16577392965630448489noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-28888161121343755302016-05-13T23:57:32.933-04:002016-05-13T23:57:32.933-04:00We are running face detector every few frames.
...We are running face detector every few frames. <br /><br /><br />Wondering how this would behave when one or more subjects tracked leave the scene and few others enter the field of view of camera Not sure if running just a face detector suffices in this case. <br /><br />Can you please confirm if ConfidenceLevel is good indicator of effective tracking when the subject just moves within the frame. Should we be looking at some other indicators. <br /><br />Would a delimiter defining the perimeter of the frame for effective tracking help? i.e., dont track beyond a predefined boundary or something similar.<br /><br /><br />Drew Sunhttps://www.blogger.com/profile/11652848166078054709noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-14741317288156428932016-05-13T23:02:48.929-04:002016-05-13T23:02:48.929-04:00I would run a face detector every few frames to ma...I would run a face detector every few frames to make sure the objects are still present.Davis Kinghttps://www.blogger.com/profile/16577392965630448489noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-20390578637661378192016-05-13T22:58:55.172-04:002016-05-13T22:58:55.172-04:00In our tests, We are tracking faces of pedestrians...In our tests, We are tracking faces of pedestrians. They walk in front of camera and move out of view. Can you please clarify what you mean by additional machenry. <br /><br />Thanks for the quick response.Drew Sunhttps://www.blogger.com/profile/11652848166078054709noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-14856515645128596382016-05-13T18:46:23.296-04:002016-05-13T18:46:23.296-04:00The confidence value is only loosely correlated wi...The confidence value is only loosely correlated with track breaking. To get a good estimate of track failure you need to include additional machenry, what exactly depends on your application.Davis Kinghttps://www.blogger.com/profile/16577392965630448489noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-27902186422568261672016-05-13T18:41:34.838-04:002016-05-13T18:41:34.838-04:00Hello Davis.
First off, Kudos for the good work....Hello Davis. <br /><br />First off, Kudos for the good work. <br /><br />We are noticing that the ConfidenceLevel of tracker is high even when the subject has moved out of the video frame. Can you please confirm the range of values for the tracker confidence level we should be looking at. Should we be looking at some other parameter for continued tracking.<br /><br />Thanks.Drew Sunhttps://www.blogger.com/profile/11652848166078054709noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-87041493763143181942016-03-23T11:52:43.534-04:002016-03-23T11:52:43.534-04:00Hi Davis,
are you open for custom job?I need to cr...Hi Davis,<br />are you open for custom job?I need to create an app which can track object. mike.sorochev@gmail.comAnonymoushttps://www.blogger.com/profile/15158031122084704846noreply@blogger.comtag:blogger.com,1999:blog-6061887630060661987.post-43977341921756518092015-12-02T02:35:27.679-05:002015-12-02T02:35:27.679-05:00This comment has been removed by the author.eyebieshttps://www.blogger.com/profile/08026232010559775090noreply@blogger.com