Research

(This page has not been updated for a long time)

We study how humans react to the real world and achieve knowledge, and develop computer algorithms that can be applied to practical situations. The relevant research areas include:

  • Machine learning

  • Signal processing

  • Computer vision.

Recently, I am interested in Deep Learning especially for multiple-level visual feature extraction.

We are interested in the applications such as

[Computer vision]

- Face detection

  • Good face detection dominates the performance of other face-related applications.

  • We use AdaBoost + LBP (local binary pattern) features for fast and efficient face detection.

- Facial emotion recognition

  • Movements of eyebrows, eyes, nose, and mouth provide cues for identifying the human emotion.

  • We investigate the methods for detecting face and its components, and develop algorithms for classifying the emotions.

  • Preliminary results:

Surprise

- Visual object tracking

  • Tracking a visual object from image sequences or video clips. We are interested in particle filters and Kalman filters. We have published a number of papers. The attached videos are generated by our proposed algorithm. The moving objects are successfully tracked even when there is occlusion.

- Object detection and recognition

  • Human object detection using thermal images:

[Acoustic signal processing]

- Sound separation and enhancement

  • Using machine learning algorithm, we can separate mixed sound sources from a single mixture input.

  • This demo is based on: A Maximum Likelihood Approach to Single-channel Source Separation. Gil-Jin Jang and Te-Won Lee. Journal of Machine Learning Research, Volume 4, pages 1365-1392, December 2003.

- Speech Recognition

  • Speech recognition is one of the largest challenges in machine learning. Although speech is very easy and natural interface to human, it is not that simple task to computers due to the complexity of the speech production organ and language structure. In many areas of engineering, many researchers have contributed to realize speech recognition. Recently, Google implemented voice-enabled search on their websites, and Android phones as well. Apple also unveiled an interface called Siri utilizing speech recognition.

  • Automatic captioning is enabled in Youtube. Try turning on caption in Youtube videos. Here are demos: (not from MI lab, but provided by Google)

Audio source separation

Generated by texts from papers regarding