A Comparative Study of Joint and Bolt Structures with and without Edge Detection using CNN

The paper aims to identify the kind of joints and bolts in the most efficient way possible. A comparative study has been performed with and without edge detection where a Convolution Neural Network (CNN) is used for classification of different joints and bolts in an image. The image is the input for the CNN model that is first converted to grayscale format and later applied to detect the edges. This study compares the image with and without any pre-processing techniques and the results are compared. Therefore, these classifications between the two neural networks are compared to find the best suited methodology for better results.