报告题目：Video Foreground Localization from Traditional Methods to Deep Learning
报告时间：2018年11月15日（星期四）上午9:00-12:00
报告地点：2教南二楼228学院会议室
报告主要内容
These days, detection of Visual Attention Regions (VAR), such as moving objects has become an integral part of many Computer Vision (CV) applications, viz. pattern recognition, object detection and classification, video surveillance, autonomous driving, and human-machine interaction (HMI). The moving object identification using bounding boxes has matured to the level of localizing the objects along their rigid borders and the process is called ForeGround Localization (FGL). Over the decades, many image segmentation methodologies have been well studied, devised, and extended to suit the video FGL. Despite that, still, the problem of video foreground (FG) segmentation remains an intriguing task yet appealing due to its ill-posed nature and myriad of applications.
The goal of this talks presents an overview of some research works that have been conducted by the speaker, which investigate the plausible real-time performant implementations from traditional approaches to modern-day Deep Learning (DL) models. It focuses mainly on improving existing methodologies through harnessing multimodal spatial and temporal cues for a delineated FGL.The first part of the talk is dedicated for enhancing conventional sample-based and Gaussian Mixture Model (GMM)-based video FGL using Probability Mass Function (PMF), temporal median filtering, and fusing CIEDE2000 color similarity, color distortion, and illumination measures, and picking an appropriate adaptive threshold to extract the FG pixels. The second part of the talk is organized to discuss about exploiting and improving Deep Convolutional Neural Networks (DCNN) for the problem as mentioned earlier. It includesnovel strategies such as double encoding - slow decoding feature learning, multi-view receptive field feature fusion, and incorporating spatiotemporal cues through 3D Convolutions (3D Conv), and Long-Short-Term Memory (LSTM) units. The application of the works is aimed towards developing intelligent transportation systems, surveillance and security, vision-guided robotics, and autonomous driving, among others.
报告人简介：
Thangarajah Akilan于2018年在加拿大温莎大学 (University of Windsor) 获得电子和计算机工程博士学位。自2008以来，他在电气和电子工程师协会(IEEE)工作了10年，同时他成为了IEEE温莎分部的秘书。他现在是加拿大温莎大学电气和计算机工程系的博士后研究员。他的研究兴趣包括计算机视觉、图像处理、机器学习、3D计算机视觉和深度学习。
报告时间：2018年11月15日（星期四）上午9:00-12:00
报告地点：2教南二楼228学院会议室
报告主要内容
These days, detection of Visual Attention Regions (VAR), such as moving objects has become an integral part of many Computer Vision (CV) applications, viz. pattern recognition, object detection and classification, video surveillance, autonomous driving, and human-machine interaction (HMI). The moving object identification using bounding boxes has matured to the level of localizing the objects along their rigid borders and the process is called ForeGround Localization (FGL). Over the decades, many image segmentation methodologies have been well studied, devised, and extended to suit the video FGL. Despite that, still, the problem of video foreground (FG) segmentation remains an intriguing task yet appealing due to its ill-posed nature and myriad of applications.
The goal of this talks presents an overview of some research works that have been conducted by the speaker, which investigate the plausible real-time performant implementations from traditional approaches to modern-day Deep Learning (DL) models. It focuses mainly on improving existing methodologies through harnessing multimodal spatial and temporal cues for a delineated FGL.The first part of the talk is dedicated for enhancing conventional sample-based and Gaussian Mixture Model (GMM)-based video FGL using Probability Mass Function (PMF), temporal median filtering, and fusing CIEDE2000 color similarity, color distortion, and illumination measures, and picking an appropriate adaptive threshold to extract the FG pixels. The second part of the talk is organized to discuss about exploiting and improving Deep Convolutional Neural Networks (DCNN) for the problem as mentioned earlier. It includesnovel strategies such as double encoding - slow decoding feature learning, multi-view receptive field feature fusion, and incorporating spatiotemporal cues through 3D Convolutions (3D Conv), and Long-Short-Term Memory (LSTM) units. The application of the works is aimed towards developing intelligent transportation systems, surveillance and security, vision-guided robotics, and autonomous driving, among others.
报告人简介：
Thangarajah Akilan于2018年在加拿大温莎大学 (University of Windsor) 获得电子和计算机工程博士学位。自2008以来，他在电气和电子工程师协会(IEEE)工作了10年，同时他成为了IEEE温莎分部的秘书。他现在是加拿大温莎大学电气和计算机工程系的博士后研究员。他的研究兴趣包括计算机视觉、图像处理、机器学习、3D计算机视觉和深度学习。