Alohomora
In this work, I presented the simplified version of Probability of Boundary (pb lite) detection algorithm, and built Convectional Neural Network(CNN) algorithm for Image classification. Results of pb lite are compared with Canny, and Sobel edge detection algorithms. Overall proved that pb lite outperforms the Sobel and Canny for edge detection. To train and test CNNs , I used CIFAR-10 dataset , and explored different methods such as ResNet, ResNeXt and DenseNet to improve the classification accuracy
PB Lite :
Report for this project can be downloaded here.
Instructor: Nitin Sanket