Analysis of fish classification system using Convolutional Neural Network (CNN) for different types of activation functions

Mansur, Mohd Aszman Syah (2022) Analysis of fish classification system using Convolutional Neural Network (CNN) for different types of activation functions. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

	Text (24 Pages) Analysis of fish classification system using Convolutional Neural Network (CNN) for different types of activation functions.pdf - Submitted Version Download (485kB)
	Text (Full text) Analysis of fish classification system using Convolutional Neural Network (CNN) for different types of activation functions.pdf - Submitted Version Restricted to Repository staff only Download (1MB)

Abstract

This project is about the classification of several types of fish by using the Convolutional Neural Network (CNN) with help of TensorFlow and Phyton. The coding process will be done by using the Google Colaboratory website. CNN is a form of artificial neural network that can be used in image recognition and processing. It has various types of activation functions, and this project will involve three types of activation functions, namely ReLU, tanh, and ELU. The dataset for this project will be obtained from the Fish4Knowledge website. The fish pictures will go through the pre-processing process for image height and width fixed resizing. After the process, the fish images will undergo a process of classification by species using CNN with three types of activation functions, namely ReLU, tanh, and ELU. Lastly, observations and comparisons will be conducted to determine which activation function is more accurate and saves time for fish classification according to the specified species.

Actions (login required)

View Item