# I need help for my program

**URL:** <https://forum.opencv.org/t/i-need-help-for-my-program/2432>\
**Category:** Python\
**Tags:** dnn, openpose\
**Created:** [March 28, 2021, 1:20pm UTC](https://forum.opencv.org/t/i-need-help-for-my-program/2432 "2021-03-28T13:20:24Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![Nytheia](https://avatars.discourse-cdn.com/v4/letter/n/9f8e36/32.png) [@Nytheia](https://forum.opencv.org/u/Nytheia)\
**Post date:** [March 28, 2021, 1:20pm UTC](https://forum.opencv.org/t/i-need-help-for-my-program/2432/1 "2021-03-28T13:20:24Z")

</div>

I want to build a program which can do detect body key points and compare to other video. I guess i need descriptor for key points and matcher for compare. But i don’t know which descriptor and matcher better for me.

\*\*I apologize if i did anything wrong. I am writing for the first time in this forum

---

<div class="post-metadata">

**Author:** ![berak](https://avatars.discourse-cdn.com/v4/letter/b/85f322/32.png) [@berak](https://forum.opencv.org/u/berak)\
**Post date:** [March 28, 2021, 1:34pm UTC](https://forum.opencv.org/t/i-need-help-for-my-program/2432/2 "2021-03-28T13:34:48Z")

</div>

you can do nthat using opencv’s dnn module.

we have a nice (even python!) [sample here](https://github.com/opencv/opencv/blob/master/samples/dnn/openpose.py)  
using either the COCO model(18 parts): [http://posefs1.perception.cs.cmu.edu/OpenPose/models/pose/coco/pose\_iter\_440000.caffemodel](http://posefs1.perception.cs.cmu.edu/OpenPose/models/pose/coco/pose_iter_440000.caffemodel)  
and [https://raw.githubusercontent.com/opencv/opencv\_extra/master/testdata/dnn/openpose\_pose\_coco.prototxt](https://raw.githubusercontent.com/opencv/opencv_extra/master/testdata/dnn/openpose_pose_coco.prototxt)

or the MPI model(16 parts):  
[http://posefs1.perception.cs.cmu.edu/OpenPose/models/pose/mpi/pose\_iter\_160000.caffemodel](http://posefs1.perception.cs.cmu.edu/OpenPose/models/pose/mpi/pose_iter_160000.caffemodel)  
and  
[https://raw.githubusercontent.com/opencv/opencv\_extra/master/testdata/dnn/openpose\_pose\_mpi\_faster\_4\_stages.prototxt](https://raw.githubusercontent.com/opencv/opencv_extra/master/testdata/dnn/openpose_pose_mpi_faster_4_stages.prototxt)

it will output landmarks like :

 ![pose2](https://us1.discourse-cdn.com/flex020/uploads/opencv/original/2X/a/a8c9dabbcc8c58d0685c7ab270599fdbf48cbba9.png)