YU-TING CHANG

Algorithms/Data/Fashion&Art

Profile


Team Member

Hello! I'm Yu-Ting Chang.

I receieved my B.S. degree in ECE Department at Taiwan National Chiao Tung University in 2015. I am currently working as a research assistant at Multimedia Computing Lab, Academia Sinica, Taiwan.

My research covers the field of Machine Learning, Multimedia, and Computer Vision. I focus on developing neural networks in the application of fashion and art related research, specialized in model combination, such as CNN based joint-training model and generative adversarial networks.

Latest CV

Research Interest

Computer Vision

I have strong multimedia cumputing background and I am especially interested in image editing techniques.

Machine Learning

I am interested in ML algorithms. I especially focus on building deep learning architectures by Keras, Tensorflow.

Fashion&Art AI

I love fashion and art! I am insterested in image retrieval, recommendation system, and style transfer techniques.

Publication

1. Yu-Ting Chang, Wen-Huang Cheng, Kai-Lung Hua, and Bo Wu, "Fashion World Map: Understanding Cities Through Streetwear Fashion," The 25th ACM International Conference on Multimedia (MM 2017), 23-27 October, 2017, Mountain View, USA.
2. Yu-Ting Chang, Min-Jhih Huang, Jorga Hu Yu, Cheng-Chun Hsu, Kai-Lung Hua, and Wen-Huang Cheng, "Fashion Eye: Understanding Streetwear Fashion Style," ChinaMM2017, Nanjing, China, September 2017.

  • Congratulations!

    Apr. 2017

    My work was accepted as a full paper by the 25th ACM Multimedia!

    I'm going to attend 2017 ACM Multimedia Conference @ Mountain View, CA, USA

    Online Paper

  • Congratulations!

    Sep. 2017

    I'm going to present Fashion World Map demo system at 2017 China Multimedia Conference (ChinaMM) @ Guangzhou, China

    Demo Website

TAKE YOUR BROKEN HEARTMAKE IT INTO ART

Project

#1

Fashion World Map: Understanding Cities Through Streetwear Fashion

Published at 2017 ACM Multimedia


  • Portfolio Item

    Intro

    A novel framework for depicting the street fashion of a city by discovering fashion items in a particular look that are most iconic for the city.

    Portfolio Item

    Global Street Fashion Dataset

    I collected Global Fashion Dataset which contains over 400K street fashion pictures and their user provided associated data.

    Portfolio Item

    Approach

    We aim to explore the iconic item of a city based on deep neural networks and PCST algorithm.

  • Portfolio Item

    Fashion World Map

    We create a fashion world map which shows the iconic street fashion of global major cities. It shows how the visual impression of local fashion cultures across the world can be depicted, modeled, and analyzed.

    Portfolio Item

    Insights

    We provided insights and observations according to our result. We can infer the Correlation of Clothing Color and City Latitude, Influence of International Clothing Brands on Local Fashion.

    Portfolio Item

    Online Demo

    A developed system to discover the city components from the user’s outfit. The application can profile the fashion taste of the users and offer smart insights embedded within the picture.

Be fearless in the pursuitof what sets your soul on fire

Activity


  • Portfolio Item

    2017 ACM Multimedia

    Oct. 2017 @ California, USA

    Portfolio Item

    2017 ChinaMM

    Sep. 2017 @ Guangzhou, China

    Portfolio Item

    2017 Taiwan HCI Workshop

    Nov. 2017 @ Taipei, Taiwan

Subscribe to stay in the loop

Drop me a line

Address

No.128, Sec.2, Academia Rd., Taipei 115, Taiwan

Phone

+886 910-927-449

Email

facelisten@gmail.com