Identifikasi Ikan Laut dengan Menggunakan Model Deep Learning Convolutional Neural Network

Carrie, Melvin (2020) Identifikasi Ikan Laut dengan Menggunakan Model Deep Learning Convolutional Neural Network. Undergraduate thesis, Universitas Internasional Batam.

[img]
Preview
Text
s-1621008-abstract-en.pdf

Download (1MB) | Preview
[img]
Preview
Text
s-1621008-abstract-id.pdf

Download (1MB) | Preview
[img]
Preview
Text
s-1621008-bibliography.pdf

Download (1MB) | Preview
[img]
Preview
Text
s-1621008-chapter1.pdf

Download (1MB) | Preview
[img]
Preview
Text
s-1621008-chapter2.pdf

Download (2MB) | Preview
[img] Text
s-1621008-chapter3.pdf
Restricted to Repository staff only

Download (251kB) | Request a copy
[img] Text
s-1621008-chapter4.pdf
Restricted to Repository staff only

Download (882kB) | Request a copy
[img]
Preview
Text
s-1621008-chapter5.pdf

Download (1MB) | Preview
[img]
Preview
Text
s-1621008-cover-id.pdf

Download (189kB) | Preview

Abstract

Convolutional Neural Network (CNN) merupakan suatu model yang dikembangkan oleh Multilayer Perceptron dan juga salah satu jenis Deep Neural Network yang dioperasikan dalam citra. Dalam penggunaan model CNN, akan diimplementasikan dalam RoV mengenai pendeteksian ikan di bawah laut. Adanya penerapan sistem tersebut, maka masyarakat yang berkehidupan sebagai perikanan maupun nelayan dapat terbantu dengan proses matapencaharain yang dilakukan sehari-hari. Dalam pengolahan CNN untuk mengindentifikasi ikan dilaut, terdapat beberapa step layer, diantaranya konvolusi layer, max pooling, flatten, dropout dan fully connected layer, yang dirancang untuk menganalisa suatu data per-step dalam struktur logika. Dengan adanya penggunaan model CNN, maka terdapat library sebagai penunjang prosesnya CNN yang terdiri dari TensorFlow, Keras dan Pillow yang digunakan untuk mempermudah dalam menjalankan suatu sistem mengindetifikasi ikan. Hasil pengujian menunjukkan algoritma dapat mendeteksi adanya ikan dari arah kiri, kanan maupun atas. Hasil pengujian yang dilakukan secara real time dengan menggunakan camera pi, menghasilkan tangkapan gambar dan informasi di shell dengan notif “fish” saat terdeteksi adanya ikan dalam laut, dan jika saat camera pi tidak mendeteksi ikan di dalam laut, maka sebaliknya akan meberikan notif “underwater”. ********************************************************************** Convolutional Neural Network (CNN) is a model developed by Multilayer Perceptron and also one type of Deep Neural Network that is operated in images. In using the CNN model, it will be implemented in RoV regarding the detection of fish under the sea. The existence of the application of this system, the people who live as fisheries and fishermen can be helped by the daily livelihood process. In CNN processing to identify fish at sea, there are several step layers, including convolution layer, max pooling, flatten, dropout and fully connected layer, which are designed to analyze a per-step data in the logical structure. With the use of the CNN model, there is a library to support the CNN process which consists of TensorFlow, Keras and Pillow which is used to make it easier to run a system of identifying fish. The test results show the algorithm can detect fish from the left, right or top. The results of tests carried out in real time using camera pi, produce images and information in the shell with the notification "fish" when detected the presence of fish in the sea, and if the camera pi does not detect fish in the sea, the reverse will give you an "underwater" notification.

Item Type: Thesis (Undergraduate)
Additional Information: Similarity: 10
Uncontrolled Keywords: CNN, RoV, Deep Learning¸ TensorFlow, Keras, Raspberry Pi, Layer, Fish
Subjects: T Technology
Divisions: School of Industrial Technology > Electrical Engineering
SWORD Depositor: Admin Repository Universitas Internasional Batam
Depositing User: Admin Repository Universitas Internasional Batam
Date Deposited: 29 Apr 2020 03:27
Last Modified: 29 Apr 2020 03:28
URI: http://repository.uib.ac.id/id/eprint/2428

Actions (login required)

View Item View Item