Light

tamasino52 / u-conddgcn Goto Github PK

Unofficial pytorch implementation of U-CondDGCN from "WenBo Hu, Changgong Zhang, Fangneng Zhan, Lei Zhang, Tien-Tsin Wong : Conditional Directed Graph Convolution for 3D Human Pose Estimation"

License: MIT License

Python 100.00%

pytorch deep-learning pose-estimation human36m 3d-pose-estimation

u-conddgcn's Introduction

Git Status

SNS

Blog / Resume

Education

(2015.03 ~ 2020.08) Bachelor's degree in Soongsil University, Seoul, KR (Dept. of Software) - Full tuition (Summa Cum Laude)
(2020.08 ~ 2022.08) Master's degree in Soongsil University, Seoul, KR (Dept. of Software)

u-conddgcn's People

Contributors

Stargazers

Watchers

Forkers

hendriktpl zdforient seongchan-jeong hanjook trellixvulnteam spinachr avotraran

u-conddgcn's Issues

about actual metris of unofficial implement.

MPJEP=43.4 ? really?

There are too many bugs in the code !!!!

The size of tensor a (7) must match the size of tensor b (8) at non-singleton dimension 2

The size of x_d2 doesn't match with x_u2. Could you please say the reason?

Node updating

U-CondDGCN/model/module/st_gcn.py

Line 552 in e87a8f8

 x = torch.stack([torch.matmul(e[:, 0], self.A[0, 1:]), x[:, 1], torch.matmul(e[:, 2], self.A[2, 1:])], dim=1) 

hi,
Are there bugs in incoming edge torch.matmul(e[ : , 0], self.A[0, 1: ]) ?
Edge (0, 1), (0, 7), (0, 4) are from the hip of a human body, if using self.A[0, 1: ], the first row of A is omitted.
From the point of view of incoming, Vertices 1, 4, 7 would get the edge from vertex 0.
suppose that torch.matmul(e[ : , 0], B)
B = [0 I]
B is 16 x 17. I is 16 x 16 identity. 0 is 16 x 1 zero column.

Reproduced result doesn't match. Could you please specify the right training procedure?

I'm getting the result below after training and testing.

===Action=== ==MPJPE===
Directions 116.01
Discussion 120.08
Eating 151.63
Greeting 120.19
Phoning 143.64
Photo 149.71
Posing 113.48
Purchases 136.89
Sitting 191.45
SittingDown 212.96
Smoking 135.34
Waiting 128.29
WalkDog 133.10
Walking 106.61
WalkTogether 110.99
Average 138.02
mpjpe: 138.02

Could you please specify the right training procedure? Do I need to pass any additional arguments?

[FAQ] How to export joint node postitions and quaternions

It would be great to add some instructions about 'How to export joint node postitions and quaternions' easily.

About training results !

Did you reproduce the accuracy reported in the paper

Recommend Projects

React

A declarative, efficient, and flexible JavaScript library for building user interfaces.
Vue.js

🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript

TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
TensorFlow

An Open Source Machine Learning Framework for Everyone
Django

The Web framework for perfectionists with deadlines.
Laravel

A PHP framework for web artisans
D3

Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

javascript

JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
web

Some thing interesting about web. New door for the world.
server

A server is a program made to process requests and deliver data to clients.
Machine learning

Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Visualization

Some thing interesting about visualization, use data art
Game

Some thing interesting about game, make everyone happy.

Recommend Org

Facebook

We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft

Open source projects and samples from Microsoft.
Google

Google ❤️ Open Source for everyone.
Alibaba

Alibaba Open Source for everyone
D3

Data-Driven Documents codes.
Tencent

China tencent open source team.