Evaluating the Efficiency of Edge Detection Algorithms for Object Classification in Convolutional Neural Networks

This Study is about comparative analysis of the efficiency of three most common edge detection algorithms which are Laplacian, Sobel and Canny edge detection. The key factor of this research lies in its resource-efficient approach, contrasting with regular method for training on a full spectrum of 255 colors. Instead, the CNN is trained on the shades of two colors: black and white. The study compares the training and testing efficiency of all the three algorithms. CNN Sequential model is used as a base to compare the training and testing efficiency of the model.