How Pose Estimation works : OpenPose Part 2

[@zuofengliu5552]

Thanks man, really great explanation

[@vandanpandya9587]

Thanks for the great explanation!

[@neuralnexus47]

very informative, thanks

[@sathyakumarn7619]

I suggest you add the author and the paper link otherwise which this is a infringement of copyrights! Thanks for the video! Go deeper in the next case

[@rajatshah125]

Thank you.

[@Rednivrug]

Thank you for the comment

[@alessandrozuech61]

Good video! Thank you!

[@Rednivrug]

Thanks Alessandro..

[@angelospapadopoulos7679]

amazing video. But if we want to implement this on our own dataset, how could we annotate this raw images? So the training input is an rgb image and a json file with the keypoints and then apply cnn regression on that?

[@prajwalshenoy9117]

Wonderful explanation. Could you share the code/implementation please?

[@Rednivrug]

Thanks @prajwal. Openpose have one integration with the opencv. You can check it out there.
github.com/opencv/opencv/blob/master/samples/dnn/openpose.py

[@akshatgovekar3646]

Angel (farishta) in disguise
People like you will definitely postpone Kalki's reanimation

[@WHAJONNALAGADDAHEMASREE]

Can we find distance between two node points for abnormal gait.? please make a video on that.

[@Darkev77]

Thanks for the great video! Question: at 5:35 why are there 38 different pairs in total? Wouldn't the number of different pairs depend on the number of humans in the scene?

[@gurvindersingh-tz1jk]

38 different pairs are for each person, the total number of pairs equal to 38*n, where n is the number of humans.

[@Darkev77]

@@gurvindersingh-tz1jk Hey thank you so much for your clarification! That makes more sense. Though could you please tell me how did we get 38 different paths? I understand that we have 18 joints and that we're trying to combine 2 joints to get a legit pair, but where did the number 38 come from? Are we calculating each pair twice (eg: neck to torso and torso to neck?) Thank you!

[@beladianeeral]

such a clear explanation. well done, paaji!

[@emmanuelkoupoh7979]

hi, thanks for your videos.... i have a question please, i want to know how encodage using part of affinity is done.

[@Rednivrug]

okay I will make the video on how encoding using part of affinity done in details

[@AdityaSingh-kp9tj]

Thanks for the video, can you make a video on PoseNet vs OpenPose .

[@Rednivrug]

Currently I'm not working on Pose Estimation Problem but in future i will try

[@AdityaSingh-kp9tj]

@@Rednivrug sure, thank you

[@haideriqbal332]

Sir i need to generate or locate auscultation points

[@esraamohamed5601]

Thank you for your helpful video , could you give me the presentation slides of this video please

[@Rednivrug]

Here is the presentation slides.
www.linkedin.com/posts/rednivrug_open-pose-presentation-activity-6580330239241613312-vkm5

[@esraamohamed5601]

@@Rednivrug Thank you !

[@vajrapuullas3265]

Ya .This is good.Can u send me the code for this.

[@gerda-morozova]

I have a question. Why VGG-19 is using there? For detecting person or what?

[@Rednivrug]

Vgg19 is kond of backbone which is used to extract the features of the image.

[@gerda-morozova]

@@Rednivrug Ok. Thanks for your answer!

[@RaselAhmed-ix5ee]

where can i get the code