SPATIOTEMPORAL MODELING FOR CROWD COUNTING IN VIDEOS

MPhil Thesis Defence


Title: "SPATIOTEMPORAL MODELING FOR CROWD COUNTING IN VIDEOS"

By

Mr. Feng XIONG


Abstract

Crowd counting is an important task in computer vision. In this thesis, we 
focus on Region of Interest (ROI) crowd counting. ROI crowd counting can be 
formulated as a regression problem of learning a mapping from an image or a 
video frame to a crowd density map. Recently, convolutional neural network 
(CNN) models have achieved promising results for crowd counting. However, even 
when dealing with video data, CNN-based methods still consider each video frame 
independently, ignoring the strong temporal correlation between neighboring 
frames. To exploit the otherwise very useful temporal information in video 
sequences, we propose a variant of a recent deep learning model called 
convolutional LSTM (ConvLSTM) for crowd counting. Unlike the previous CNNbased 
methods, our method fully captures both spatial and temporal dependencies. 
Furthermore, we extend the ConvLSTM model to a bidirectional ConvLSTM model 
which can access long-range information in both directions. Extensive 
experiments using publicly available datasets demonstrate the reliability of 
our approach and the effectiveness of incorporating temporal information to 
boost the accuracy of crowd counting. In addition, we also explore transfer 
learning for crowd counting. Our transfer learning experiments show that once 
our model is trained on one dataset, its learning experience can be transferred 
easily to a new dataset which consists of only very few video frames for model 
adaptation. At last, we also introduce the application of our methods in 
practical project.


Date:			Thursday, 17 August 2017

Time:			2:00pm - 4:00pm

Venue:			Room 2610
 			Lifts 31/32

Committee Members:	Prof. Dit-Yan Yeung (Supervisor)
 			Prof. Albert Chung (Chairperson)
 			Dr. Ming Liu (ECE)


**** ALL are Welcome ****