List

We have been working on content analysis (machine intelligence over large-scale image/video streams) since 2002 as having my PhD in Columbia Univ. Yes, in the old days, we used no deep learning but hand-crafted features. It’s so tough for dealing with such “beast” signals.

Even in the old days, we were working on “large-scale” concepts (e.g., LSCOM [1]), which was huge at that time (2004) but only had 449 concepts (classifiers). We delivered one of the leading video research and detection system in NIST TRECVID benchmarks (e.g., [2]) over the “largest” dataset ( ~250 hours, at that time).

It’s so exciting to see the huge improvement in the past years, especially, triggered by the availability of quality training data, huge computation resources, and the advent of machine learning techniques.

These days, during my sabbatical visit to IBM Watson Research, I am so excited to witness and contribute the “beauty” — the things we had strived for for years have already turned live in life (e.g., the first-ever AI movie trailer [3], detecting the live golf TV streams for the highlights [4]).

We did learn a lot in this winding trajectory from the mistakes, the setbacks, and the excitements. However, I do feel lucky and grateful that we believe in and continue with what we have been doing. I can project that there will be much more exciting moments to come!!!

[1] M. Naphade, J. R. Smith, J. Tesic, S.-F. Chang, W. Hsu, L. Kennedy, A. Hauptmann, J. Curtis, “Large-Scale Concept Ontology for Multimedia, ” IEEE Multimedia Magazine, 13(3), 2006.

[2] Shih-Fu Chang, Wei Jiang, Winston Hsu, Lyndon Kennedy, Dong Xu, Akira Yanagawa, Eric Zavesky, “Columbia University TRECVID-2006 Video Search and High-Level Feature Extraction,” in NIST TRECVID workshop, Gaithersburg, MD, Nov. 2006.

[3] “IBM Watson creates the first AI-made film trailer – and it’s incredibly creepy,” http://www.wired.co.uk/article/ibm-watson-ai-film-trailer

[4] https://twitter.com/IBMSports/status/851256590816735232

 

  Posts

1 2 3 6
June 7th, 2018

First Place (#1) in Disguised Face Recognition in CVPR 2018

January 6th, 2019

意想不到的科技部「AI投資潛力獎第一名」

December 16th, 2018

Keep Recruiting for Machine Learning Research Partners for Numerous Visual Sensors

December 16th, 2018

FutureTech Demo and Breakthrough Award (未來科技突破獎)

October 12th, 2018

結合虛與實的試鞋生成網路 (Virtual Try-On Shoe with Generative Neural Networks)

我們都有這樣的困擾,在電子購物的時候,看到一雙好看的鞋子,想買。但是卻又拿不定主意自己穿起來好看嗎?或是搭配某件褲子適合嗎?怎麼讓網路虛擬商城的鞋子,可以有效試在自己的腳上呢? 這個工作的挑戰在於如何使用單張鞋子商品的照片,很自然的合成在使用者的腳上,而且腳可能會有各種姿勢、角度。如何客服這個問題? 很高興大學部專題生(EE) 周晁德 完成了這個 PIVTONS 的虛擬鞋子試穿生成網路,試著解決這個困擾大家很久的問題。 這個有趣的工作也將於十二月初,在澳洲珀斯舉辦 Asian Conference on Computer Vision (ACCV) 2018 以大會演說 (Oral) 的方式跟大家分享這個工作。接下來全新的測試資料集將會公開讓大家使用,如果可以的話,我們也將試試看將整個試穿生成系統上線,讓大家體驗虛擬試鞋的樂趣 — 可以多試穿,多省錢。 我們鼓勵high-risk的研究工作。令人慶幸的是,這工作的發想、資料收集都是專題生獨立完成。當然在過程當中遇到很多GAN生成的問題,網路設計、訓練的問題,幾乎放棄了,還好團隊成員一起想辦法解決,關關難過,關關過(甚至免費擔任model),讓這個兼具技術深度以及商業價值的系統,可以順利完成。 我們也一直努力,讓智能生成(或是辨識)系統,賦予更有意義的應用 […]

September 13th, 2018

Finalist (Top 3) in 2018 IEEE Signal Processing Society Video and Image Processing (VIP) Cup

July 29th, 2018

信手拈來的3D模型搜尋 (Cross-View and Cross-Domain 3D Model Search)

July 27th, 2018

低解析人臉辨識跟解析度放大 (Very Low-Resolution Face Hallucination and Recognition)

June 18th, 2018

Winning Third Place in CVPR 2018 Video Recognition Challenge — Moments in Time

June 13th, 2018

[Video Report] National Investment for the GPU Supercomputer?

June 9th, 2018

Amazing Crowd Size and Positive Feedbacks in the Deep Learning Lecture for GTC 2018 Taipei