EFFECTS OF SEGMENTATION AND SCALING ON ANOMALY DETECTION USING CONVOLUTIONAL NEURAL NETWORKS

Anomaly detection have been a successful technique in real world applications and mostly in industrial applications where precision is to be valued the most. Images are a day-to-day part of our lives, so here I have tried to present an approach for solving the issue of anomaly detection using Convolutional Neural Networks. The Neural network has been trained on diverse images of both anomalous images and Good (Normal) image. Neural Networks have been an integral and also emerged as a effective Deep Learning approach for tasks like image classification, anomaly detection and many such classification tasks. In recent years convolutional neural networks have been emerged as favorites for image classification tasks because of its various functionalities. The implications of this research are to present the effects of Scaling and Edge Preservation, Segmentation on images gives us different results of similar Neural network and to choose best of them. Thus, experimenting the Convolutional Neural Networks on different parameters yields us various results.