Explainable Machine Learning Models and Architectures

2Artificial Intelligence and Image Recognition Algorithms

Siddharth¹*, Anuranjana¹ and Sanmukh Kaur²

¹Dept. of Information Technology, Amity School of Engineering and Technology, Amity University Uttar Pradesh, Noida, India

²Dept. of Electronics and Communication Engineering, Amity School of Engineering and Technology, Amity University Uttar Pradesh, Noida, India

Abstract

Various computer vision tasks, including video object tracking and image detection/classification, have been successfully achieved using feature detectors and descriptors. In various phases of the detection-description pipeline, many techniques use picture gradients to characterize local image structures. In recent times, convolutional neural networks (CNNs) have taken the place of some or all these algorithms that reply and detectors and descriptors. Earlier algorithms, such as the Harris Corner Detector, SIFT, ASIFT and SURF that were once hailed as cutting-edge image recognition algorithms have now been replaced by something more robust. This document highlights the generational improvement in image recognition algorithms and draws a contrast between current CNN-based image recognition techniques and previous state-of-the-art algorithms that don’t rely on CNN. The purpose of this study is to emphasize the necessity of using CNN-based algorithms over traditional algorithms which are not as dynamic.

Keywords: Convolution Neural Networks (CNN), Scale Invariant Feature Transform (SIFT), Artificial Intelligence ...