An Improved Deep Convolutional Neural Network (DCNN) for finding the Fish Freshness

Abstract

Finding the fish freshness is most significant task with existing models. It is very important to overcome the various issues such as accuracy in detecting the fish freshness. Deep Learning (DL) is most widely used to detect the process of finding the solutions to the complex tasks. This paper mainly focused on detects the fish freshness automatically by analyzing the fish images. In this paper, A Shallow deep Convolutional Neural Network (SD-CNN) is proposed in this work instead of device approaches. A pre-trained model which is VGG-16 architecture is used to extract the features of the fish images to find the freshness of the fish or not. In order to classify fish images, an improved classifier is used to create drop-out and dense layers. Then a novel deep learning algorithm is applied on the fish dataset. Different methods are used to find the fish freshness such as SVM, DCNN. By using these methods several drawbacks are identified such as lack of accuracy, precision and Recall. The proposed model is SD-CNN that improves the performance in terms of accuracy is 97.76%, Precision is 97.06 and recall is 0.96.