Install npm and hexo

1
2
3

$ sudo apt update
$ sudo apt install nodejs npm
$ npm install -g hexo-cli

Install and update next theme

See hexo github.

1
2
3

# Installation
$ cd [Blog_Folder]
$ git clone https://github.com/theme-next/hexo-theme-next themes/next

1
2
3

# Update
$ cd [Blog_Folder]/themes/next
$ git pull

Hexo commands

See Official Documentation.

1
2
3
4
5
6
7
8
9
10

# New post
$ hexo new <title> 
# Clean `publish` folder
$ hexo clean 
# Generates static files
$ hexo generate # or hexo g
# Starts a local server. By default, this is at http://localhost:4000/
$ hexo server # or hexo s
# Deploy the `publish` folder to github
$ hexo deploy # or hexo d

Personalization of next theme

In themes/next/_config.yml, I selected the two-column Pisces theme, enabled social icons, added rounded avatar of myself, set text-align to left, and changed the icon between year and copyright info to a groom. There are also other minor changes which are not listed below.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25

scheme: Pisces
...
# Icon between year and copyright info.
  icon:
    # Icon name in Font Awesome. See: https://fontawesome.com/icons
    name: fa fa-broom
    # If you want to animate the icon, set it to true.
    animated: false
    # Change the color of icon, using Hex Code.
    # color: "#ff0000"
...
# Sidebar Avatar
avatar:
  # Replace the default image and set the url here.
  url: /images/me_compressed.jpg
  # If true, the avatar will be dispalyed in circle.
  rounded: true
  # If true, the avatar will be rotated with the cursor.
  rotated: false
...
social:
  Zurich: https://www.google.ch/maps/place/Z%C3%BCrich/@47.2088177,8.1869486,7z/data=!4m2!3m1!1s0x47900b9749bea219:0xe66e8df1e71fdc03 || fas fa-location-arrow
  GitHub: https://github.com/Adamink || fab fa-github
  E-Mail: mailto:wuxiao_1997@outlook.com || fa fa-envelope
  Resume: https://drive.google.com/file/d/1SPAb0IV08PyrDARwEW2kmDtMD4VnQIXG/view?usp=sharing || fas fa-file

Also, the default page width of Pisces is too wide for me, so I edited themes/next/source/css/variables/Pisces.styl:

1
2
3
4
5

$content-desktop              = 'calc(100% - %s)' % unit($content-desktop-padding / 2, 'px');
// $content-desktop-large        = 1160px;
// $content-desktop-largest      = 73%;
$content-desktop-large   = 800px;
$content-desktop-largest = 50%;

Link a custom domain name to Github Pages

The default domain name of github pages will be [***].github.io. We can replace it with custom domain name.

1. Change records at DNS server.
   
2. Add CNAME to source folder, so that it will be deployed every time we call hexo deploy. The contents of CNAME file is simply the domain name.

   1	wuxiao.site

Useful links:

• How do I link my domain to GitHub Pages
• Managing a custom domain for your GitHub Pages site

This repository is a course project of Physically-Based Simulation in Computer Graphics, ETH Zurich, Fall 2020.

The project is a collaborative work from Ge Cao, Xiao Wu, Mingyang Song.

The demo video is available at Youtube.

Features:

• Particle-based fluid simulation with SPH method
• Iterative SESPH
• Coupling liquid simulation with rigid body (the wheel)
• Dambreak scene & Waterwheel scene
• Particles import & export in ply and cfg format
• Multiple ways of visualization using OVITO, OpenGL/GLUT and blender
• Surface reconstruction using splashsurf
• Multithreading for accelerating simulation using OpenMP
• Rendered with GPU accelerated ray-tracing using Cycles

Install & Run

The installation has only been tested on Ubuntu for now. The project is built with CMake. Before installation, please make sure CMake is installed. The project also depends on OpenGL/GLUT, please use the following command to install dependencies.

1

sudo apt-get install libglu1-mesa-dev freeglut3-dev mesa-common-dev libx11-dev libxi-dev

To build the project, run the following command in the project folder.

1
2
3
4

mkdir build
cd build
cmake ..
make

Then, type ./SphWaterSimulation to run the simulation.

Render

For further rendering, we use splashsurf to reconstruct the liquid surface, and use blender to render the scene. When importing sequence of .obj files into blender, please refer to plugin stop motion obj for blender.

To install relevant packages, please refer to scripts/install_utilities.sh

After installation, use python3 ./scripts/construct_surface.py to construct liquid surface.

We also upload our blender file to Google Drive for future reference Link.

Config

The config file is SphWaterSimulation/constants.h. To disable visualization, change variable IF_VISUALIZE to false. To disable rigid body(the wheel), change kUseRigidBody to false.

[PDF]

The implementation of Multipath Event-Based Network for Low-Power Human Action Recognition (WFIoT 2020).

Introduction

Event-based cameras are bio-inspired sensors capturing asynchronous per-pixel brightness changes (events), which have the advantages of high temporal resolution and low power consumption compared with traditional frame-based cameras.

We propose a multipath deep neural network for action recognition
based on event camera outputs. Extensive experiments verify the effectiveness of the proposed model with a recognition accuracy of 85.91% on the DHP19 dataset.

Installation

This repository uses PyTorch, which can be installed by following commands.

1

conda install pytorch torchvision torchaudio cudatoolkit=10.2 -c pytorch

Dataset

Offical Dataset

The offical DHP19 dataset is avaiable here.

Preprocessing

After downloading, Matlab scripts are used to pre-process the event data as in matlab/DHP19/generate_DHP19/Generate_DHP19.m​. This script will generate a bunch of event data ended with .h5​​. Then src/dataset/pose7500.py​ is used to turn .h5​ files into .npy​ files for faster dataset load.