International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 10 Issue: 07 | July 2023
p-ISSN: 2395-0072
www.irjet.net
Automated Fish Species Detection R. Kranthi Kumar1, D Ashritha2, D Siddharth3, D Neeraj Ashish4, M Akshara5 1Assistant Professor, VNR Vignana Jyothi Institute of Engineering & Technology, Hyderabad
2345Under Graduate Student, VNR Vignana Jyothi Institute of Engineering & Technology, Hyderabad
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Fish species detection is essential for many
disease management strategies. It also aids in ensuring proper fish health and welfare, preventing the spread of invasive species, and facilitating the breeding and genetic improvement of specific fish species with desirable traits
different uses, including managing fisheries and monitoring aquatic ecosystems. In this research, we present a deep learning-based method for precise fish species identification using MobileNetV2 architecture. The MobileNetV2 is a compact and effective convolutional neural network (CNN) model that successfully strikes a compromise between precision and computational effectiveness, making it especially ideal for contexts with limited resources. We assembled a large collection of fish photos from various species with different lighting, backdrops, and orientations in order to evaluate our method. We demonstrate the efficiency of our approach in attaining accurate fish species detection and categorization through comprehensive training and evaluation. Notably, our methodology maintains computational economy while delivering competitive performance compared to cutting-edge technologies, enabling its use in real-time scenarios. By presenting an automated system for fish species detection, our proposed application aims to streamline and enhance the efficiency of monitoring aquatic ecosystems. This, in turn, contributes significantly to the conservation of biodiversity by providing a more precise and efficient means of assessing and managing fish populations.
Fish species detection involves the use of various techniques and technologies to identify and classify different fish species based on their physical characteristics, genetic markers, or behavioral patterns. These methods can range from traditional visual identification by experienced aquaculturists to advanced technologies like DNA analysis, computer vision, and machine learning algorithms. This research provides an extensive examination of the latest techniques and advancements in automated fish species detection. Our focus is specifically directed toward exploring the utilization of the MobileNetV2 architecture. MobileNetV2 is an efficient and lightweight convolutional neural network (CNN) model that effectively balances accuracy and computational efficiency. The primary objective of this paper is to provide researchers, practitioners, and enthusiasts with a comprehensive understanding of the current methodologies, challenges, and future directions in automated fish species detection using MobileNetV2. We also discuss the underlying principles, network structure, and transfer learning strategies employed in adapting MobileNetV2 for this specific application.
Key Words: (Fish species detection, MobileNetV2, CNN, Aquatic ecosystems.)
1.INTRODUCTION Fish species detection plays a crucial role in the field of aquaculture, which is the cultivation of aquatic organisms such as fish, shellfish, and aquatic plants for various purposes. The essence of aquaculture lies in its ability to meet the increasing global demand for seafood, alleviate pressure on wild fish stocks, and promote sustainable food production.
Through this research paper, we aim to provide researchers and practitioners with a valuable resource that can guide future developments in automated fish species detection, foster collaborations, and promote advancements in this critical field of study.
2. LITERATURE SURVEY
In order to effectively manage aquaculture operations and ensure optimal growth and health of fish populations, it is essential to accurately identify and monitor the different fish species present. This is particularly important in situations where multiple species coexist within the same aquatic environment, such as in fish farms or natural water bodies used for aquaculture purposes.
2.1. Research on fish identification in tropical waters under unconstrained environment based on transfer learning[2022]: The paper focuses on fish detection in tropical waters using transfer learning in an unrestricted environment. The method begins with pre-processing of the images using affine transformation to improve the data. Then, a RestNet50 deep convolutional neural network is built using transfer learning. The effectiveness of fish detection and recognition is compared prior to and following
Accurate species detection in aquaculture has several benefits. It enables aquaculturists to maintain speciesspecific environmental conditions, feed formulations, and
© 2023, IRJET
|
Impact Factor value: 8.226
|
ISO 9001:2008 Certified Journal
|
Page 500