Master thesis presentation by Linnéa Claesson and Björn Hansson: Deep Learning Methods and Applications:  Classification of Traffic Signs and Detection of Alzheimer’s Disease from Images

Chalmers logo

Chalmers logo

Welcome to the master thesis presentation by Linnéa Claesson and Björn Hansson: Deep Learning Methods and Applications:  Classification of Traffic Signs and Detection of Alzheimer’s Disease from ImagesPresented by Linnéa Claesson and Björn Hansson.  
 
When? Friday, January 27, 14:00
Where? Landahlsrummet (room 7430), Hörsalsvägen 11, 7th floor, Chalmers
 
Examiner: Irene Gu
 
Abstract
In this thesis work, Linnéa Claesson and Björn Hansson investigate deep learning methods for two classification problems, namely traffic sign recognition and Alzheimer’s disease detection, through convolutional neural networks (CNNs). The two datasets used are from the German Traffic Sign Recognition Benchmark (GTSRB) and the Alzheimer’s Disease Neuroimaging Initiative (ADNI). For the GTSRB dataset, our final test results have reached above 97 % classification accuracy on 43 different signs. We have also tested different parameter settings under the selected CNN structure, and their impact to the classification accuracy. For binary classification of MRI and DTI images from the ADNI dataset consisting of healthy brains and brains afflicted with Alzheimer’s disease, detection of Alzheimer’s disease has yielded about 65 % accuracy during testing. These results show that CNNs are very promising for classifying traffic signs, however, further investigation needs to be done when dealing with the more complicated problem of detecting Alzheimer’s disease.