Identification and Classification of Industrial Elements using Artificial Intelligence and Image Processing Techniques

This paper presents a model which identifies various industrial elements with the help of Image Classification algorithms and the concepts of image processing. The proposed model provides a solution where the industry is hazardous and wants to evolve in automation field to make the process safer and easier. The performance analysis of popular Convolutional Neural Network is implemented in order to identify the objects of a plant. The concepts of Deep Learning Algorithms are applied with the help of Tensor Flow software and Keras API. As first step the CNN architecture was decided manually. After analyzing the results, the next step is implementation of Transfer Learning techniques to improve the accuracy. Some of the most powerful Transfer Learning algorithms are being used like ResNet, Inception, Mobile Net, DenseNet, VGG16 and Xception etc. Each of these algorithms is evaluated and the output accuracy is tested. The dataset is collected by taking problem statement into consideration. The models are verified and the experimental results exhibited an accuracy of almost 90%.