Introduction

• A place to 'post' the progress of my master's thesis.

Progress(Related studies necessary for master 's thesis.)

• Where can I get the data?

• https://accad.osu.edu/research/mocap/mocap_data.htm

• http://mocapdata.com/

• what is BVH file?

  

• Reading BVH file?

• Since the skeleton information of 'ACCAD' and MOCAPDATA.COM data is different, each preprocessing is necessary.(ACCAD dataset -> Used in Progress , MOCAPDATA.COM dataset -> used in Master's Thesis)

• Convert from C3D to BVH using MotionBuilder - It can be difficult because it uses a professional program called 'Motion Builder'.

• Code

• PreProcessing the data Using Motionbuilder - Click on this Sentence and Learn

• Writing BVH file?

• Code

• Human Motion Prediction basic -Various attempts

• Motion Prediction advanced Seq2Seq batch learning version - completed

• Motion Prediction Seq2Seq - Not Using Decoder Input - completed

• Motion Prediction Simple - completed

• Motion Prediction encoding decoding - completed

• Motion Prediction encoding decoding using BidirectionalCell - completed

• Motion Prediction encoding decoding : Joint angle to Cartesian coordinates using open source - Fixing

• Motion Prediction decoding encoding : using training_set and test_set - completed

• Motion Prediction basic Seq2Seq sequencial learning version - completed

• Motion Prediction advanced Seq2Seq sequencial learning version - completed

• You have to learn one by one from easy data.(Please refer to the code.)

• You have to learn one by one from easy data.(Please refer to the code.)

• You have to learn one by one from easy data.(Please refer to the code.)

• You have to learn one by one from easy data.(Please refer to the code.)

• You have to learn one by one from easy data.(Please refer to the code.)

• You have to learn one by one from easy data.(Please refer to the code.)

• Encoding + Decoding Structure

• Encoding + Decoding Sequence to Sequence Structure

Master's Thesis

• Human motion for Master's Thesis

• 1. Motion Classification

• 2. Human Motion Pattern Classification

• 3. Motion Prediction(Generation) With more data

  Question? Can the network characterize motion data? Is it possible to generalize the Human motion?

  Development environment

• 1-1. Motion Prediction(Generation) , Batch Learning , Input with noise , Tied weights , Training method = Test method

• 1-2. Motion Prediction(Generation) , Batch Learning , Input with noise , Tied weights , Training method != Test method

  '''Please read the notes below'''* about Network structure<span class="o">-</span> <span class="n">If</span> <span class="n">we</span> <span class="n">share</span> <span class="n">encoder</span> <span class="ow">and</span> <span class="n">decoder</span> <span class="n">weights</span><span class="p">,</span> <span class="n">we</span> <span class="n">can</span> <span class="n">only</span> <span class="n">use</span> <span class="n">Residual</span> <span class="n">Connection</span> <span class="k">for</span> <span class="s">&#39;One RNN Layer&#39;</span><span class="p">,</span> 
  <span class="n">because</span> <span class="s">&#39;ResidualCell&#39;</span> <span class="n">must</span> <span class="n">have</span> <span class="n">the</span> <span class="n">same</span> <span class="n">size</span> <span class="n">of</span> <span class="nb">input</span> <span class="ow">and</span> <span class="n">output</span><span class="o">.</span>
  
  <span class="o">-</span> <span class="n">The</span> <span class="n">code</span> <span class="k">for</span> <span class="n">MultiLayer</span> <span class="ow">is</span> <span class="ow">not</span> <span class="n">actually</span> <span class="n">used</span><span class="p">,</span> <span class="n">but</span> <span class="ow">is</span> <span class="n">written</span> <span class="k">for</span> <span class="n">later</span> <span class="n">use</span><span class="o">.</span>* about Dataset<span class="o">-</span> <span class="o">&lt;</span><span class="n">Training</span> <span class="n">data</span><span class="o">&gt;</span> <span class="p">:</span> <span class="n">The</span> <span class="n">training</span> <span class="n">motion</span> <span class="n">data</span> <span class="p">(</span><span class="mi">272</span><span class="p">)</span> <span class="k">for</span> <span class="mi">4</span> <span class="n">women</span><span class="p">,</span> <span class="mi">68</span> <span class="n">motion</span> <span class="n">data</span> <span class="n">per</span> <span class="n">person</span><span class="o">.</span>
  
      - <Test data> : The Test motion data for 1 women, 68 motion data per person.

• Human Motion Analysis consists of 3-Projects

  - <Training data> : The training motion data (1224) for 10 men and 8 women, 68 motion data per person.- <Test data> : The Test motion data (204) for 2 men and 1 women, 68 motion data per person.

• window 10.1 64 bit and Ubuntu Linux 16.04.2 LTS

• python verison : 3.6.1 , anaconda3 version : (4.4.0)

• pycharm Community Edition 2017.2.2

Dependencies

• mxnet-0.12.1(window) , mxnet-0.12.1(Linux)

• tqdm -> (progress) , graphviz -> ( Visualization )

链接：