An Examination of Image Identification and Classification through the Implementation of a Hybrid Deep Neural Network (LSTM-CNN)
Punyaban Patel, G. Madhukar, Edem Suresh Babu, Soma Kalyani · 2024
This article describes how a basic Convolutional Neural Network (CNN) is incorporated using Long Short-Term Memory (LSTM), creating an entirely novel approach in the well-studied area of picture categorization. A type of recurring neural network (RNN) with the ability to remember dependence over time is called LSTM. It has been noted that when employed in a stacked sequence, LSTMs can enhance CNN's capacity for extraction of features. For extended periods, LSTMs can retain patterns carefully, and CNNs can extract significant characteristics from them. When employed for recognizing images, this layered LSTM-CNN structure has an advantage beyond traditional CNN classifiers. The suggested model is durable and appropriate for a variety of issues related to classification because it is built on collections of artificial neural networks, like recurring and convolutional neural networks. We executed our hypothesis on two typical data sets in order to confirm our findings. The significance of our proposed model has been established by comparing the findings with those of other classifications.