News

MCL Segmentation Research Work to be Presented at ICCV 2015

The segmentation research in our lab made a significant breakthrough these days. The paper “Robust Image Segmentation Using Contour-guided Color Palettes” by Xiang Fu, Chien-Yi Wang, Chen Chen, Changhu Wang and C.-C. Jay Kuo was accepted by ICCV 2015. In this paper, the contour-guided color palette (CCP) is proposed to efficiently integrate contour and color cues of an image. To find representative colors of an image, color samples along long contours between regions, similar in spirit to machine learning methodology that focus on samples near decision boundaries, are collected to achieve an image-dependent color palette. This color palette provides a preliminary segmentation in the spatial domain, which is further fine-tuned by post-processing techniques such as leakage avoidance, fake boundary removal, and small region mergence. While CCP offers an acceptable standalone segmentation result, it can be further integrated into the framework of layered spectral segmentation to produce a more robust segmentation.

 

For more details, please refer to our paper that will be published soon. The latest code of this paper can be downloaded here. It is written in MATLAB, and has been tested under 64-bit Windows, Linux, and Mac OSX.

By |September 20th, 2015|News|Comments Off on MCL Segmentation Research Work to be Presented at ICCV 2015|

Interview with new MCL member He Ming Zhang

In 2015 Fall semester, MCLab has a new PhD student, He Ming Zhang. She received the B.S. degree in Communication Engineering and M.S degree in , both from Delft University of Technology, Delft, Netherlands. Now we have an interview with her, talking about her background and her thoughts on USC and MCLab.

1. Could you briefly introduce yourself? 

I received my Bachelor of Science in Electrical Engineering at Delft University of Technology (TU Delft). After that, I continued my study at Delft for a master degree in multimedia signal processing. My research experiences include designing protocols for secure signal processing and analyzing algorithms for distributed signal processing.

2. What is your first impression of USC and MCL?

Unlike TU Delft, USC has more diversity in schools and also the students. The first time I walked through the campus, I was impressed by the high density of buildings.

MCL has a really large number of students but Prof. Kuo still manages to work quite close with students. It is amazing that I can meet him at least twice per week. The whole group looks like a big family. People communicate with each other very well.

3.What’s your future expectation for MCL?

I hope I will soon know everyone well and become good friends with them. I desire for a friendly study and research environment where people are inspired by each other. I also hope that I can enjoy the research here and contribute to the field of computer vision.

By |September 14th, 2015|News|Comments Off on Interview with new MCL member He Ming Zhang|
  • Permalink Gallery

    Congratulations to two MCL PhD students for passing their defense

Congratulations to two MCL PhD students for passing their defense

Congratulations to Sanjay Purushotham and Pang-Chang (Brian) Lan, who passed their defense last week! Following are their thesis abstracts and they also shared their PhD experiences.
Thesis: Advanced Machine Learning Techniques for Video, Social, and Biomedical Data Analytics (Sanjay Purushotham)
In this thesis, advanced machine learning techniques are developed to tackle challenging problems arising in three Big Data application domains. They are: 1) partial near-duplicate content copy detection and alignment for the multimedia (video) application, 2) personalized single user and group recommender systems for the social media data application, and 3) sparse learning models for identification of discriminative feature interactions for gene expression prediction and cancer stage classification for the biomedical data application. Novel and suitable machine learning algorithms and models are designed to meet the nature of the data in each specific application domain. 

Thesis: Secure Wireless Communications with Side Information: Secrecy Analysis and Unitary Modulation Realization (Brian Lan)
In wireless communication systems such as LTE, information security is protected by symmetric cryptography which assumes that the user and the base station shares the same secret key. But before the setup of the symmetric cryptosystems, the key sharing and authentication processes are vulnerable to eavesdropping. This is referred to as the secure initiation problem. Our work proposes to use physical layer techniques to achieve perfect secrecy without using cryptography, so the secure initiation problem can be solved. In particular, we bring up the concept of having only channel state information (CSI) at the transmitter side but not at the receiver nor the eavesdropper. This concept will be shown beneficial in secrecy comparing to the conventional assumption of full CSI at all terminals. A practical scheme, unitary modulation, is then proposed to exploit this concept. [...]

By |September 6th, 2015|News|Comments Off on Congratulations to two MCL PhD students for passing their defense|
  • Permalink Gallery

    Congratulations to three MCL PhD students for passing their defense

Congratulations to three MCL PhD students for passing their defense

Congratulations to Harshad Kadu, Joe Yuchieh Lin and Xiang Fu, who passed their defense last week! Following are their thesis abstracts and they also shared their PhD experiences.

Thesis: Advanced Techniques for Human Action Classification and Text Localization (Harshad Kadu)

The thesis contains two main research topics:

1) Automatic human action classification with mocap data

We propose a TSVQ based multi-resolution string representation scheme that transforms the time-series of human poses into codeword sequences. Temporal and spatial features extracted from these sequences are combined together using novel fusion methods to achieve superior performance.

2) Text localization in natural scene images

Stable extremal region operator detects regions of interest in the proposed multi-stage incremental region classification framework. Geometric filtering, context-based text grouping and ensemble classifier stages, remove false positives and group related text regions into words.

 

Thesis: Experimental Design and Evaluation Methodology for Human-Centric Visual Quality Assessment (Joe Yuchieh Lin)
The problem of human-centric visual quality assessment (VQA) is extensively studied in this thesis. Our study includes three major topics: 1) design of a dataset for streaming video quality assessment, 2) development of a new and effective video quality assessment index, 3) exploration of a new methodology for human visual quality assessment based on the notion of just-noticeable-differences (JND).
 

Thesis: Advanced Visual Segmentation Techniques: Algorithm Design and Performance Analysis (Xiang Fu)

Two research topics are covered in the dissertation:

1) How to interactively represent and segment the object(s) in a video?

An interactive video object segmentation framework is proposed to handle complex and diverse scenes with large motion and occlusions, which outperforms the state-of-the-art Adobe After Effect.

2) How to reliably generate automatic image segmentation?

Region-Dependent Spectral Graph is designed to fuse contour, surface, and depth cues according to the type of regions. Depth cue can further merge the regions in the textured area, which is never applied for segmentation before.

Contour-guided Color [...]

By |August 31st, 2015|News|Comments Off on Congratulations to three MCL PhD students for passing their defense|

Interview with new MCL member Yuhang Song

MCL has a new Ph.D student, Yuhang Song in Fall 2015. Let’s give him a warm welcome!

Yuhang received B.S. degree from the EE Department of Tsinghua University in Fall, 2015. He decided to join USC MCL to pursue his Ph.D degree beginning from this Fall semester. We had a brief interview with him.

What is your first impression of USC and MCL?

USC is a beautiful campus with high cultural diversity, and MCL is a big group with friendly mates. It really impressed me a lot during the first group meeting, where team members talked about their current work. Their enthusiasm about research and technology led a strong influence on me.

Could you briefly introduce yourself? (Previous research experience, project experience, research interest and expertise)

I had a couple of research experience about multiple fields, including pedestrian action categorization, image retrieval, structural light coding and time series analysis.  After my attendance of this lab, I want to keep an open mind on a variety of topics of computer vision and machine learning, and to dive into certain interesting topic after several tries. 

What’s your future expectation for MCL?

As I’m new here, I hope to learn more about the ongoing projects and state-of -the-art technology in the lab. After that, I hope to devote myself to developing my research abilities, as well as communication skills. It will be a great journey in my life during this period in USC and MCL, and I’ll do my best to be a good team member and keep good relationship with my colleagues.

By |August 23rd, 2015|News|Comments Off on Interview with new MCL member Yuhang Song|

Interview with new MCL member Hsin-Ying Lee

MCL has a new Ph.D student, Hsin-Ying Lee in Fall 2015. Let’s give Hsin-Ying a warm welcome!

Hsin-Ying received B.S. degree from the EE Department of National Taiwan University in 2014 Spring. He got the USC-Taiwan Fellowship award and decided to join USC MCL to pursue his Ph.D degree beginning from this Fall semester. We had a brief interview with him.

What is your first impression of USC and MCL?

USC is quite an impassioned and multicultural place. People in MCL are full of enthusiasms for cutting-edge research.

Could you briefly introduce yourself? (Previous research experience, project experience, research interest and expertise)

My previous research field was electronic design automation (EDA). I led a team to win two international computer-aided design (CAD) contests. I had joined a MediaTek project about datapath extraction and placement. Now I would like to dive into computer vision and start a new journey of research. Currently, my research interests are biomedical related subjects, depth estimation from single image, etc..

What’s your future expectation for MCL?

Because I am just a beginner in computer vision, I have a lot to learn. I hope I can develop some advanced and useful solutions for computer vision problem. On the other hand, I am looking forward to getting to know all team members in the lab and becoming friends with all of you. It is my honor to work with so many brilliant colleagues and I hope I will be a good teammate as well.

By |August 16th, 2015|News|Comments Off on Interview with new MCL member Hsin-Ying Lee|
  • Permalink Gallery

    NVIDIA Supports MCL in Building CNN for Computer Vision Research

NVIDIA Supports MCL in Building CNN for Computer Vision Research

MCL is joining the research of Convolutional Neural Network (CNN). With the donation of one K40 GPU from NVIDIA, MCL will create its first GPU server to train and test various CNN architectures. K40 is the prestigious GPU that is widely used in parallel computing environment. It features 12 GB memory and 2880 CUDA cores that can significantly accelerate the speed of modern computer vision algorithms. When ECC is turned off and clock frequency is slightly boosted, it is capable of processing 20 iterations of Caffe training in 19.2 seconds.

MCL director, Prof. C.-C. Jay Kuo, appreciates NVIDIA’s generous donation and has assigned Phd students, Hao Xu and Qin Huang to build a powerful server to house the K40. Prof. C.-C. Jay Kuo is interested in solving various computer vision problems. He has a vision that there should be a systematic way to provide unified solution to all types of computer vision problems. To achieve that goal, Prof. C.-C. Jay Kuo believes that we need to better understand the strength of the automatically trained feature sets.

MCL has a detailed road map for its future development of Convolutional Neural Network, and one crucial step is to study the features trained through it. MCL will use Caffe and Theano libraries to test various CNN architectures, from the classic AlexNet to the deeper and more accurate VGG net. The visualization of features trained from these architectures will be carefully examined using deConv network. Finally, MCL wants to build a clear understanding towards both the automatically trained features themselves and the reason for why are they selected by the CNN.

By |August 9th, 2015|News|Comments Off on NVIDIA Supports MCL in Building CNN for Computer Vision Research|
  • Permalink Gallery

    MCL interns and students presented their case studies in the entrepreneurship short course

MCL interns and students presented their case studies in the entrepreneurship short course

Three MCL students presented their case studies about business startups in industry in the last lecture of the entrepreneurship course this summer.

Chien-Yi Wang presented Adobe. Adobe has historically focused upon the creation of multimedia and creativity software products, with a more recent trend towards rich Internet application software development, including digital marketing and digital media solutions. The company is famous for its development on the “Portable Document Format”, which is adopted worldwide as a common medium for electronic documents. After the hallmark technology of PDF standard, they were devoted to develop powerful software and services for people to create suitable digital content, deploy it across media and devices, measure and optimize it over time and achieve greater business success. Adobe has a unique business model which could optimize the revenue by distributing their products in several ways. The company spent lots of money on the marketing side which allowed them to stand out among their competitors. Nowadays, their products all go on the cloud for customer to subscribe and make people easier to collaborate and use via different platforms. Adobe is also leading computer vision researches which could benefit the whole digital media community.

Eddy Wu presented Fitbit Inc. Fitbit is one of the most well-known technology companies that produce activity trackers – the wireless wearable devices that measure data such as number of steps, quality of sleep, and other personal metrics. Fitbit is a fast growing company since its establishment; its revenue increased from $14.5 million in 2011 to $745 million in 2014. Fitbit went public in Jun, 2015, and the shares jumped to about $30 a piece at IPO day and valued the company over $6 billion. There are many reasons for its huge success. First of all, while its competitors put more efforts on the hardware development, [...]

By |August 2nd, 2015|News|Comments Off on MCL interns and students presented their case studies in the entrepreneurship short course|
  • Permalink Gallery

    MCL interns and students presented their case studies in the entrepreneurship short course – II

MCL interns and students presented their case studies in the entrepreneurship short course – II

In the last lecture of the entrepreneurship short course, another three MCL interns presented their case studies on typical business startups.

Yifang Chen presented Zhihu, a Chinese Q & A website launched in 2011. Zhihu was similar to Quora at its initial stage but later developed its own specialty. She mainly discussed Zhihu’s (as a Social Network Service (SNS) Website) focus on knowledge and intelligence sharing. Her presentation started with how “Cognitive Surplus” phenomena laid the foundation for the Q&A platforms. Then she specifically analyzed the different requirements of techniques to meet the changes in Zhihu’s business strategy, which also changed: useful content, people and core values. Through her presentation, audience did not only learn about how Zhihu accumulated resource, built up its own brand and stepped into expansion but also realize the gradual topic change, or even the deviation off the website’s initial purposes. Her presentation was interactive and interesting to all audiences.

Melody Zhou presented Airbnb which created an excellent solution for home renters and providers worldwide. Since its inception in 2008, Airbnb had started an innovative community marketplace for people to post, discover, and book a wide variety of accommodations around the world. Airbnb has pioneered a new industry of “collaborative consumption” and peer-to-peer accommodation rentals, which leaves room for the company’s potential growth and worldwide adoption. Melody mainly talked about the history of Airbnb, how it was established and several key ingredients of its rapid expansion among many start-ups. Also, she shared some personal experience as a customer. Finally Melody pointed several current challenges for Airbnb, its reaction and future plan. Melody made an objective assessment. Her view was highly valued by the audiences.

Shanglin Yang presented Tesla Motors. The Tesla Motors is the leading company in electric cars, electricvehicle powertrain components, and battery products. Nowadays, Tesla has attracted more and more attention for its success of Model S product, especially in the area of electric vehicle, which is in general hard for new company to [...]

By |July 26th, 2015|News|Comments Off on MCL interns and students presented their case studies in the entrepreneurship short course – II|
  • Permalink Gallery

    Three MCL interns presented their case studies in the entrepreneurship short course

Three MCL interns presented their case studies in the entrepreneurship short course

In the last lecture of the entrepreneurship short course, three MCL interns presented their final projects – case studies on successful companies. Their successful and impressive presentations showcased remarkable amount of time and effort they had devoted to the projects. By finishing these projects, the interns acquired fundamental business knowledge and rich experience on technology and business innovations in the MCL.

Kevin Wang presented Zappos, a thriving online shoe and clothing store. Zappos was founded in 1999 by Nick Swinmurn and obtained their first investment funds of $500,000 from Tony Hsieh of Ventures Frog, who joined Zappos one year later, serving as the Co-CEO. According to Kevin, what distinguishes Zappos from other companies is its core value, “Deliver Happiness”. For its employees, Zappos creates a happy and family-like work environment. For its customers, Zappos wants to “wow” the customer so that he or she will come back. Knowing how to retain its customers is what keeps Zappos profitable. Each of its employees needs to go through a four-month training to improve their capability in serving customers and implementing company culture. It’s no wonder that Zappos has an estimated 75% customer retention rate.  In 2014, Zappos faced a major change in the way it was structured in 2014. It began its transition from a traditional corporate structure to a holacracy, a new type of corporate structure designed to improve communication flow.

Matthew Trans presented Riot Games, an unquestionable world top-tier game development company despite of the fact that it only has one game: League of Legends. Even though League of Legends is a free game, Riot still managed to attain revenue of $624 million last year just from microtransactions. Riot’s success roots deeply in its philosophy of creating the most-player-focused game in history. At Riot, players come first, and revenue second. Not only does Riot’s priority [...]

By |July 19th, 2015|News|Comments Off on Three MCL interns presented their case studies in the entrepreneurship short course|