Welcome to the Mulimedia Computing Group!!
The main research topics addressed by our group are in the field of the “Mulimedia". Here are some reseach topics:
Deep Learning is a set of algorithms in machine learning that attempt to model high-level abstractions in data by using architectures composed of multiple non-linear transformations. Theoretical results suggest that in order to learn the kind of complicated functions that can represent high-level abstractions(e.g., in vision, language, and other AI-level tasks), one may need deep architectures. Various deep learning architectures such as deep neural networks, convolutional deep neural networks, and deep belief networks have been applied to fields like computer vision, automatic speech recognition, natural language processing, and music/audio signal recognition where they have been shown to produce state-of-the-art results on various tasks. Our group has some related research about deep learning including theoretical improvements and applications. Now we are concentrating on dropout in deep learning and multi-task deep learning.
Query Difficulty Estimation (QDE) or Query performance prediction (QPP) attempts to quantify the quality of the search results returned for a given query without relevance judgments or user feedback. This technique is significant for the following reasons: 1) Feedback to the user: Users can rephrase a “difficult” query to improve system effectiveness. 2) Feedback to the search system: The search system can invoke alternative retrieval strategies for “difficult” queries according to their estimated difficulty. 3) Feedback to the system administrator: The administrator can identify queries related to a specific subject that are “difficult” for the search engine, and expand the collection to better answer poorly covered subjects. 4) For distributed information retrieval: Estimation can be used to decide which search engine to use by estimating the results of which search engine are best. Additionally, estimation can be used as a method for merging the results of queries performed on different datasets by weighting the results from each dataset.
Image Aesthetic Quality Assessment has long been an attractive computer vision research topic. It aims at classifying images into “good” or “bad” according to their beauty. Another direction is to give a specific aesthetic score or ranking. This work can be inserted into many other systems and gain amazing applications. Take content-based image retrieval for instance, with effective aesthetic quality assessment algorithms, it can return both relevant and beautiful images to the consumers and therefore achieve better user experience. Also for the photo sharing websites or offline users, this research provide them a new photo management style.
Local Features have proven to be successful in application such as matching, object recognition, texture recognition, image retrieval, video data mining and recognition of object categories. They are distinctive, robust to occlusion, and do not require segmentation. Recent work has concentrated on making these descriptors invariant to image transformations. We need to detect image regions covariant to a class of transformations, which are then used as support regions to compute invariant descriptors. Our work focus on building a proper local feature which is robust to changes. In this way we can improve the retrieval system and the recognition system at the same time.
Recommendation. With the concept of WEB 2.0 becoming popular in late 2004,
Human Activity Recognition is an important area of computer vision research today. The goal of human activity recognition is to automatically analyze ongoing activities from an unknown video (i.e. a sequence of image frames). In a simple case where a video is segmented to contain only one execution of a human activity, the objective of the system is to correctly classify the video into its activity category. In more general cases, the continuous recognition of human activities must be performed by detecting starting and ending times of all occurring activities from an input video. The ability to recognize complex human activities from videos enables the construction of several important applications. Automated surveillance systems in public places like airports and subway stations require detection of abnormal and suspicious activities, as opposed to normal activities. For instance, an airport surveillance system must be able to automatically recognize suspicious activities like “a person leaving a bag” or “a person placing his/her bag in a trash bin.” Recognition of human activities also enables the real-time monitoring of patients, children, and elderly persons. The construction of gesture-based human computer interfaces and vision-based intelligent environments becomes possible with an activity recognition system as well.
Do you want to join us?
News and Events
May. 2015: One paper gets accepted by Neurocomputing.
Apr. 2015: One paper gets accepted by Multimedia Tools and Applications.
Apr. 2015: One paper gets accepted by IJCAI.
Mar. 2015: One paper gets accepted by ICME.
Jan. 2015: One paper gets accepted by Pattern Recognition.