科研项目

项目列表


Computer Vision Meets Cognitive Science


计算机视觉与认知科学


Introduction

 

Computer vision has achieved significant progress in detecting, tracking and recognizing objects in real images. However, beyond the scope of traditional question of “what is where” in the theory framework of David Marr, it lacks the abilities to understand scenes characterizing human visual experience. We try to understand scenes from the perspective of computer vision and cognitive science.

 

 

What is a container?

 

Containers are ubiquitous in daily life. By container, we consider any physical object that can contain other objects, such as bowls, bottles, baskets, trashcans, refrigerators, etc. We are interested in the following questions: What is a container? Will an object contain another object? How many objects will a container hold? We study those problems by evaluating human cognition of containers and containing relations with physical simulation. In the experiments, we analyze human judgments with respect to results of physical simulation under different scenarios. We conclude that the physical simulation is a good approximation to the human cognition of container and containing relations.

 

Video

 

 

Papers

 

1. Wei Liang, Yibiao Zhao, Yixin Zhu, and Song-Chun Zhu. "Evaluating Human Cognition of Containing Relations with Physical Simulation." CogSci, 2015. [PDF]

 

 

Head Pose Estimation

 

We propose a method to estimate head pose with convolutional neural network, which is trained on synthetic head images. We formulate head pose estimation as a regression problem. A convolutional neural network is trained to learn head features and solve the regression problem. To provide annotated head poses in the training process, we generate a realistic head pose dataset by rendering techniques, in which we consider the variation of gender, age, race and expression. Our dataset includes 74000 head poses rendered from 37 head models. For each head pose, RGB image and annotated pose parameters are given. We evaluate our method on both synthetic and real data. The experiments show that our method improves the accuracy of head pose estimation.

 

Video

 

 

 

Papers

 

1. Xiabing Liu, Wei Liang, Yumeng Wang, Shuyang Li, and Mingtao Pei. " 3D Head Pose Estimation with Convolutional Neural Network Trained on Synthetic Images." ICIP, 2016. [PDF]

 

 

Game Design and Programming

 

We focus on serious game design and programming, which is mainly related to children education and human study. As a part of training program, we also realize some novel and interesting games. In the realization of these games, we develop broad cooperation with professional designers to inspire and motivate our creativity.

 

Video

 

 

Papers