Abstract:In recent times, marine fish species recognition becomes an important research area to protect the ocean environment. It is a tough and time-consuming operation to manually detect marine fish species on the ocean floor. Depending on the situation, extensive sample efforts may be required. These efforts might be harmful to the marine ecosystem. Automated classification methods are capable of properly classifying these fish on a consistent basis. An increasing number of people are becoming interested in utilizing electronic monitoring and reporting with artificial intelligence for the aim of fish identification and enhancing present techniques. It is becoming more usual to use video and pictures of fish (either underwater or on ships) in fishing operations. These techniques are operational, transportable, and non-invasive, and they provide high-quality pictures at a lower cost than traditional approaches. Automated image processing techniques such as Deep Learning (DL) and Machine Learning (ML) are now available, and they may be customized to perform efficient fish species identification and segmentation. In this aspect, this paper presents an Intelligent DL based Marine Fish Species Classification (IDL-MFSC) technique. The proposed IDL-MFSC technique involves three major processes such as pre-processing, fish detection and fish classification. Primarily, Weiner filtering-based noise removal process takes place as a pre-processing step. In addition, Mask R-CNN (Mask Region Based Convolutional Neural Networks) with Residual Network as a backbone network is used for fish detection. Moreover, Optimal Deep Kernel Extreme Learning Machine (ODKELM) based classification method is employed for determining the class labels of the marine fish species in which the parameter tuning of the DKELM model takes place using Water Wave Optimization (WWO) technique. The performance of the proposed method is tested using an openly accessible Fish4Knowledge dataset. The experimental result highlights the supremacy of the IDL-MFSC technique over the recent techniques with respect to various measures.

Underwater Fish Species Classification using Convolutional Neural Network and Deep Learning

Classification of Freshwater Zooplankton by Pre-trained Convolutional Neural Network in Underwater Microscopy

Fish Detection and Classification Based on Improved ViT

Fish species classification in unconstrained underwater environments based on deep learning

An Automatic Detection and Statistical Method for Underwater Fish Based on Foreground Region Convolution Network (FR-CNN)

Fish Species Classification Using Deep Learning and Appearance-Based Feature Extraction

Fish Species Image Classification Using Convolutional Neural Networks

Recognition Of Fish Categories Using Deep Learning Technique

Crosspooled FishNet: transfer learning based fish species classification model

Multi-classification deep neural networks for identification of fish species using camera captured images

Fish detection and species classification in underwater environments using deep learning with temporal information

A Deep CNN-Based Salinity and Freshwater Fish Identification and Classification Using Deep Learning and Machine Learning

Fish Recognition Using Convolutional Neural Network

Deepwater-SE: A new automatic fish classification network based on deep learning

Fish species classification using a collaborative technique of firefly algorithm and neural network

Deep but Lightweight Neural Networks for Fish Detection

Fish Species Recognition using Deep Learning Techniques

Enhanced deep learning models for automatic fish species identification in underwater imagery

Fish Detection in Underwater Environments Using Deep Learning

An Intelligent Deep Learning Enabled Marine Fish Species Detection and Classification Model

A Survey of Deep Learning Techniques for Underwater Image Classification