90 likes | 182 Views
An ANN Approach to Identify if Driver is Wearing S afety B elts. Hanwen Chen 12/9/2013. Image Samples Both Negative and Positive samples; Same angle, same resolution, pre-processed. Feature Extraction SIFT: Scale-invariant Feature Transform; HOG: Histogram of Oriented Gradients.
E N D
An ANN Approach to Identify if Driver is Wearing Safety Belts Hanwen Chen 12/9/2013
Image Samples • Both Negative and Positive samples; • Same angle, same resolution, pre-processed. • Feature Extraction • SIFT: Scale-invariant Feature Transform; • HOG: Histogram of Oriented Gradients. • Classifier (SVM) • Support Vector Machine.
Image Samples: • Positive Samples • Negative Samples
Support Vector Machine • We are familiar with that!
Collect Samples • Feature Extraction • VLFeat Open-source Library • http://www.vlfeat.org/ • Classify • Statistics Toolbox (Matlab R2013b) • Neuron Network Toolbox (Matlab R2013b)
With SIFT feature: • Confusion matrix: • Miss rate: 30% • With HOG feature: • Confusion Matrix: • Miss rate: 30%