Skip to content

OpenPose_en

Jump to bottom
Huang Jian Cong edited this page Mar 12, 2018 · 4 revisions

<< 返回主页

EN | 中文

Contents

  1. Summary
  2. Video
  3. Quickstart
  4. Structure
  5. FAQ

Summary

This is a demo that makes baxter to simultaneously identify and track many people, including the positioning of the key points of the human face, the positioning of the key points of the human body and the positioning of the various joints of the human body.If the hardware is perfect, you can track up to 500 people.

So far this node has been tested on ROS Indigo powered by Ubuntu 14.04 only. If you got any problem with the code, please contact us.

All you need first is to build OpenPose (Intalltion OpenPose in Ubuntu)

Files Location:

Video

IMAGE ALT TEXT

Quickstart

Before you run this package, you need to install the following dependencies

  • CUDA 8.0
  • cuDNN 5.1
  • OpenCV 3.1

The following is the command line for installing dependencies.

# Clone Openpose package
git clone https://github.com/CMU-Perceptual-Computing-Lab/openpose

# Update the system kernel
sudo apt-get update && sudo apt-get upgrade && sudo apt-get dist-upgrade && sudo apt-get autoremove

# Build up the dependence environment of caffe
sudo apt-get install libatlas-base-dev
sudo apt-get install libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libboost-all-dev libhdf5-serial-dev
sudo apt-get install libgflags-dev libgoogle-glog-dev liblmdb-dev protobuf-compiler
sudo apt-get install python-dev python-numpy python-scipy python-matplotlib

# Install CUDA 8.0
bash ./openpose/ubuntu/install_cuda.sh

# Install cuDNN 5.1
bash ./openpose/ubuntu/install_cudnn.sh

# Install caffe and Openpose
bash ./ubuntu/install_caffe_and_openpose_if_cuda8.sh

# Running Openpose
 - Video
./build/examples/openpose/openpose.bin --video examples/media/video.avi --face --hand

 - IMAGE
./build/examples/openpose/openpose.bin --image_dir examples/media/ --face --hand

 - On Webcam 
./build/examples/openpose/openpose.bin --face --hand

To read more detail or find the solutions to some problems, please refer to FAQ part below.

Structure

waiting...

FAQ

Q1. Besides the NVIDIA graphics card, are there other graphics CARDS available for Openpose?

A1. No, the model of openpose is trained by caffe, and Nvidia graphics card must be used.

Q2. I get an error similar to: Check failed: error == cudaSuccess (2 vs. 0) out of memory.

A2. Most probably cuDNN is not installed/enabled, the default Caffe model uses >12 GB of GPU memory, cuDNN reduces it to ~1.5 GB.

Q3. I get an error similar to: /sbin/ldconfig.real: /usr/lib/nvidia-375/libEGL.so.1 not a symbol link /sbin/ldconfig.real: /usr/lib32/nvidia-375/libEGL.so.1 not a symbol link.

A3.

sudo mv /usr/lib/nvidia-375/libEGL.so.1 /usr/lib/nvidia-375/libEGL.so.1.org

sudo mv /usr/lib32/nvidia-375/libEGL.so.1 /usr/lib32/nvidia-375/libEGL.so.1.

sudo ln -s /usr/lib/nvidia-375/libEGL.so.375.39 /usr/lib/nvidia-375/libEGL.so.1

sudo ln -s /usr/lib32/nvidia-375/libEGL.so.375.39 /usr/lib32/nvidia-375/libEGL.so.1

Q4. OpenPose is quite slow, is it normal? How can I speed it up?

A4. Check the OpenPose Benchmark to discover the approximate speed of your graphics card. Some speed tips:

1. Use cuDNN 5.1 (cuDNN 6 is ~10% slower).
2. Reduce the `--net_resolution` (e.g. to 320x176) (lower accuracy).
3. For face, reduce the `--face_net_resolution`. The resolution 320x320 usually works pretty decently.
4. Use the `MPI_4_layers` model (lower accuracy and lower number of parts).
5. Change GPU rendering by CPU rendering to get approximately +0.5 FPS (`--render_pose 1`).

https://github.com/birlrobotics

Clone this wiki locally