Shufan Yang

A neuro-inspired visual tracking method based on programmable system-on-chip platform

Using programmable system-on-chip to implement computer vision functions poses many challenges due to highly constrained resources in cost, size and power consumption. In this work, we propose a new neuro-inspired image processing model and implemented it on a system-on-chip Xilinx Z702c board. With the attractor neural network model to store the object’s contour information, we eliminate the computationally expensive steps in the curve evolution re-initialisation at every new iteration or frame. Our experimental results demonstrate that this integrated approach achieves accurate and robust object tracking, when they are partially or completely occluded in the scenes. Importantly, the system is able to process 640 by 480 videos in real-time stream with 30 frames per second using only one low-power Xilinx Zynq-7000 system-on-chip platform. This proof-of-concept work has demonstrated the advantage of incorporating neuro-inspired features in solving image processing problems during occlusion.

A Low Computational Approach for Assistive Esophageal Adenocarcinoma and Colorectal Cancer Detection

In this paper, we aim to develop a low-computational system for real-time image processing and analysis in endoscopy images for the early detection of the human esophageal adenocarcinoma and colorectal cancer. Rich statistical features are used to train an improved machine-learning algorithm. Our algorithm can achieve a real-time classification of malign and benign cancer tumours with a significantly improved detection precision compared to the classical HOG method as a reference when it is implemented on real time embedded system NVIDIA TX2 platform. Our approach can help to avoid unnecessary biopsies for patients and reduce the over diagnosis of clinically insignificant cancers in the future.

Interactive reading using low cost brain computer interfaces

This work shows the feasibility for document reader user applications using a consumer grade non-invasive BCI headset. Although Brain Computer Interface (BCI) type devices are beginning to aim at the consumer level, the level at which they can actually detect brain activity is limited. There is however progress achieved in allowing for interaction between a human and a computer when this interaction is limited to around 2 actions. We employed the Emotiv Epoc, a low-priced BCI headset, to design and build a proof-of-concept document reader system that allows users to navigate the document using this low cast BCI device. Our prototype has been implemented and evaluated with 12 participants who were trained to navigate documents using signals acquired by Emotive Epoc.

Presenting and Investigating the Efficacy of an Educational Interactive Mobile Application for British Sign Language Using Hand Gesture Detection Techniques

In this paper we present the design and development of a mobile application that assists learners of British Sign Language. The application uses interaction; namely, image recognition of a participants gestures to give feedback as to the correctness of the gestures. Through designing with different algorithms and a user test, the efficacy of such an application is presented along with limitations that are present with current technology.