About this fork

This fork contains pure python version of Realtime Multi-Person Pose Estimation. Initially it was forked from Michal Faber fork, all credit for porting original work to Keras goes to him.

I this fork I've reimplemented images argumentation in pure python, it is significanly shorter(285 lines vs 1202 lines in Michal Faber's C++ rmpe_server, and way less than in original work)

Despite of Python language this code is significantly faster than original implementation(140 images/s vs 30 images/s C++ code on my machine). This is not really useful since most of people don't have 5 GPUs, but just to prove the point python programs could be fast. The magic is in combining all affine transformations to one matrix, and calling single warpAffine, and vectorized numpy computation of PAFs and Heatmaps.

Could be run as iterator inside train_pose.py (default), or as separate ./rmpe_server.py

Current status

• image augmentation: rotate, shift, scale, crop, flip (implemented as single affine transform, i.e. much faster)
• mask calculation: rotate, shift, scale, crop, flip
• joint heatmaps
• limbs part affinity fields
• quality is same as original work and bit better than Michal's version.

Current work

• Ability to easily modify config and train different models. See addins submodule for head detector example and example how to add new datasets(MPII, Brainwash)

Realtime Multi-Person Pose Estimation

This is a keras version of project

Introduction

Code repo for reproducing 2017 CVPR paper using keras.

Results

 

Contents

1. Converting caffe model
2. Testing
3. Training

Require

1. Keras
2. Caffe - docker required if you would like to convert caffe model to keras model. You don't have to compile/install caffe on your local machine.

Converting Caffe model to Keras model

Authors of original implementation released already trained caffe model which you can use to extract weights data.

• Download caffe model cd model; sh get_caffe_model.sh
• Dump caffe layers to numpy data cd ..; docker run -v [absolute path to your keras_Realtime_Multi-Person_Pose_Estimation folder]:/workspace -it bvlc/caffe:cpu python dump_caffe_layers.py Note that docker accepts only absolute paths so you have to set the full path to the folder containing this project.
• Convert caffe model (from numpy data) to keras model python caffe_to_keras.py

Testing steps

• Convert caffe model to keras model or download already converted keras model https://www.dropbox.com/s/llpxd14is7gyj0z/model.h5
• Run the notebook demo.ipynb.
• python demo_image.py --image sample_images/ski.jpg to run the picture demo. Result will be stored in the file result.png. You can use any image file as an input.
• python demo_camera.py to run the web demo.

Training steps

UPDATE 26/10/2017

Fixed problem with the training procedure. Here are my results after training for 5 epochs = 25000 iterations (1 epoch is ~5000 batches) The loss values are quite similar as in the original training - output.txt

Results of running demo_image --image sample_images/ski.jpg --model training/weights.best.h5 with model trained only 25000 iterations. Not too bad !!! Training on my single 1070 GPU took around 10 hours.

UPDATE 22/10/2017:

Augmented samples are fetched from the server. The network never sees the same image twice which was a problem in previous approach (tool rmpe_dataset_transformer) This allows you to run augmentation locally or on separate node. You can start 2 instances, one serving training set and a second one serving validation set (on different port if locally)

• Install gsutil curl https://sdk.cloud.google.com | bash. This is a really helpful tool for downloading large datasets.
• Download the data set (~25 GB) cd dataset; sh get_dataset.sh,
• Download COCO official toolbox in dataset/coco/ .
• cd coco/PythonAPI; sudo python setup.py install to install pycocotools.
• Go to the "training" folder cd ../../../training.
• Generate masks python generate_masks.py. Note: set the parameter "mode" in generate_masks.py (validation or training)
• Create intermediate dataset python generate_hdf5.py. This tool creates a dataset in hdf5 format. The structure of this dataset is very similar to the original lmdb dataset where a sample is represented as an array: 5 x width x height (3 channels for image, 1 channel for metedata, 1 channel for miss masks) For MPI dataset there are 6 channels with additional all masks. Note: set the parameters datasets and val_size in generate_hdf5.py
• Download and compile the dataset server rmpe_dataset_server. This server generates augmented samples on the fly. Source samples are retrieved from previously generated hdf5 dataset file.
• Start training data server in the first terminal session. ./rmpe_dataset_server ../../keras_Realtime_Multi-Person_Pose_Estimation/dataset/train_dataset.h5 5555
• Start validation data server in a second terminal session. ./rmpe_dataset_server ../../keras_Realtime_Multi-Person_Pose_Estimation/dataset/val_dataset.h5 5556
• Optionally you can verify the datasets inspect_dataset.ipynb
• Set the correct number of samples within python train_pose.py - variables "train_samples = ???" and "val_samples = ???".
  This number is used by keras to determine how many samples are in 1 epoch.
• Train the model in a third terminal python train_pose.py

Related repository

• CVPR'16, Convolutional Pose Machines.
• CVPR'17, Realtime Multi-Person Pose Estimation.

Citation

Please cite the paper in your publications if it helps your research:

@InProceedings{cao2017realtime,
  title = {Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields},
  author = {Zhe Cao and Tomas Simon and Shih-En Wei and Yaser Sheikh},
  booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2017}
  }