A Deep Machine Learning Neural Network for Real Time Object Classification using Keras & Tensorflow

Abstract

Every object has certain patterns with respect to their geographical architecture. Analyzing these patterns for tracking objects at real time is effective to contribute in artificial intelligence towards automation. There are various conventional methods have been introduced till now that recognize objects on the basis of patterns, Convolutional Neural Network (CNN) is one of them. But CNN is limited with certain objects because of analogous patterns of different objects. System confuses or does not work effectively when multiple objects are intended to recognize at real time. Tensorflow is most effective object detection API for identifying multiple objects at real time with high level of precision. It is used for both research and production at Google. Here the system uses tensorflow
and Keras for machine learning along with Deep Neural Network to recognize different kind of objects in a single frame. Keras is an open source neural network library written in Python. It is capable of running on top of TensorFlow, Microsoft Cognitive Toolkit or Theano. Designed to enable fast experimentation with deep neural networks, it focuses on being user-friendly, modular, and extensible. Proposed system uses python for fine tuning the algorithm for object detection with different kind of packages such as pillow, pandas-datareader, scipy and jupyter. System is intended to classify the objects with high level of accuracy with quick response.