Human Pose Detection using Python and OpenCV - OpenPose - with code on Colab

Here we go over a human pose detection system using CMUs OpenPose System

Support me on Patreon

💻 General Tech

📸 Camera Gear

🎥Video Accessories

👨‍💻Books I like on Personal Growth

👨‍💻Books I like on Investing

DISCLOSURE: Some of the links on this page are affiliate links, meaning, at no additional cost to you, I may earn a commission if you click through and make a purchase. Affiliate commissions help fund videos like this one.

[Syzygyyy]

Congrats you're the only one that gives a colab for pose estimation that works out of the box :)

[yogeshmahajan]

One of the finest videos of misbah Mohammad

[floydmotumi]

import cv2
cap =
ret, frame = cap.read()
frame_height, frame_width, _ = frame.shape
out = cv2.VideoWriter('output.avi', cv2.VideoWriter_fourcc('M', 'J', 'P', 'G'), 10, (frame_width, frame_height))
print("Processing Video...")
while cap.isOpened():
ret, frame = cap.read()
if not ret:
out.release()
break
output = poseDetector(frame)
out.write(output)
out.release()
print("Done processing video")

Processing Video...
Done processing video

Please note the video does not play - Zero Results = No Output, please re-test and confirm if there's any change required. Thanks

[apat]

Hey Mishbah Mohammed, I am currently trying to process a video and I am getting an error saying poseDetector is not defined, and I am returned the output.avi video but it is corrupt/giving an error. Could you possible guide me as to how I can fix this, as I am fairly new at this subject. Thanks!

[yodewmg]

you told that if I want to use webcam, just remove the dance.mp4 and put the "0, 0" right? so i changed like this cap = cv2.VideoCapture("0, 0") but when i started the code, there is an error in the frame_height, frame_width, _ = frame.shape code. do you know why? i really want to know it. and how can i solve it?

[shubhammathur]

I am facing an error while read(run) the graph_opt file
net =
someone help me regard this error.

[rs]

Great work, can you make a simple video how to transfer learn train tensorflow model by adding custom dataset to the coco dataset

[yashwanthyashwanth]

How can I extract the features from the from that skeleton for gait recognition can u help me with an idea?

[RaselAhmed-ixee]

how can recognize the activity from the final pose estimation like whether the person is siiting, standing,walking,running,sleeping? please reply its very urgent!

[pythonner]

it would have been cool to use this for vr tracking

[sehulviras]

Hi,

How do you solve the problem of jitters. I want the joints to say on the body and not bounce around. Any suggestions ?

[Capanos]

Hi, in Localhost I adapted your code to make it work by inserting Friend images.
Everything works but I would like the "poses" lines to be more precise.

I read your note "I Modified The OpenCV DNN EXAMPLE TO USE etc ..."

You can give me support by telling me how to improve the dataset❤

[SHASHIKUMAR-pphg]

AttributeError Traceback (most recent call last)
in <module>()
1 input = cv.imread("ex.jpg")
----> 2 output = poseDetector(input)
3 cv2_imshow(output)

in poseDetector(frame)
23
24 def poseDetector(frame):
---> 25 frameWidth = frame.shape[1]
26 frameHeight = frame.shape[0]
27

AttributeError: 'NoneType' object has no attribute 'shape'

I am unable to Apply this on a local file please help

[snehitvaddi]

Hey Mohammed, could you tell me how to find the angle between different parts like the angle made by limbs etc...!!

[ShopperPlug]

where does the processed images are stored?

[bacem]

i want to detect human activities (walking, sleep, ...) solution!

[nusharawedasingha]

it worked well for images but does not work for video

[atharvavaidya]

Can I put fall detection in this project

[ssaniljainn]

is it possible to keep a track of movements of particular joints? Or is there a need to include datasets?
i have seen action recognition type applications based on this, although not sure where to even start

[ferosekhan]

Assalamualikum brother

Thank you very much, It is very helpful!

Why cannot execute if give silhouettes images?