Claim Missing Document
Check
Articles

Found 2 Documents
Search

Pengembangan Aplikasi Login Menggunakan Pengenalan Wajah Dan Kedipan Mata Frindi Mangimbulude; Pinrolinvic D.K. Manembu; Feisy Diane Kambey
CogITo Smart Journal Vol. 9 No. 1 (2023): Cogito Smart Journal
Publisher : Fakultas Ilmu Komputer, Universitas Klabat

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31154/cogito.v9i1.453.97-108

Abstract

Facial recognition technology is widely used as one of the technologies to authenticate someone. However, some devices do not yet have a facial recognition component that is able to recognize real people who have been recognized or only photos because they only embed a 2D facial recognition system which is less secure, especially on devices with low specifications and older devices, and there are still many people at this time who still use such devices but they also need security. So in this study a login system with facial recognition and eye blink detection was created which aims to provide more security on older devices or devices with low specifications. This study uses the Local Binary Pattern Histogram facial recognition algorithm and the Face landmark wink detection algorithm from the dlib library. So that the application can run properly on older devices or devices with low specifications.
Klasifikasi Ikan Cakalang dan Tongkol Menggunakan Convolutional Neural Network : Fish Classification of Skipjack and Mackerel Tuna Using Convolutional Neural Network Wellifan Arrank Tonapa; Pinrolinvic D.K. Manembu; Feisy D. Kambey
Jurnal Teknik Informatika Vol. 19 No. 01 (2024): Jurnal Teknik Informatika
Publisher : Universitas Sam Ratulangi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35793/jti.v19i01.52013

Abstract

Abstract — Indonesia has a rich diversity of fish species, especially marine fish species. However, the abundance of fish species also poses challenges for the community in classifying each species. This challenge becomes even more significant when dealing with species that share similar physical characteristics, such as the pelagic fish group, which includes skipjack tuna (Katsuwonus pelamis) and mackerel tuna (Euthynnus affinis). Therefore, it is essential to have a profound understanding of fisheries science to accurately classify each fish species. With advancements in current technology, species classification can be automated using image-based classification methods. This research employs the Convolutional Neural Network (CNN) method to classify skipjack tuna and mackerel tuna species. The research results in a CNN classification model constructed using a transfer learning approach by leveraging the pre-trained ResNet50 model available in Keras Applications. The CNN Classification Model generated achieves a performance with a 95% accuracy rate, an average macro precision of 95%, an average macro recall of 95%, and an average macro F1 score of 95%. Key words— Classification; Convolutional Neural Network; fish species; Transfer Learning; Image. Abstrak — Indonesia memiliki banyak keanekaragaman spesies ikan, terutama spesies ikan laut. Namun, keberagaman spesies ikan yang banyak juga menimbulkan kesulitan bagi masyarakat dalam melakukan klasifikasi pada setiap spesies ikan yang ada. Apalagi, pada beberapa spesies ikan yang memiliki fisik yang hampir sama, seperti kelompok ikan pelagis, yaitu cakalang (Katsuwonus pelamis) dan tongkol (Euthynnus affinis). Oleh karena itu, penting untuk memiliki pemahaman mendalam tentang ilmu perikanan agar dapat melakukan klasifikasi yang benar terhadap setiap spesies ikan. Dengan kemajuan teknologi saat ini, pengklasifikasian spesies ikan dapat dilakukan secara otomatis menggunakan metode klasifikasi berdasarkan citra. Penelitian ini menggunakan metode Convolutional Neural Network (CNN) untuk mengklasifikasikan spesies ikan cakalang dan tongkol. Penelitian ini menghasilkan model klasifikasi CNN yang dibangun menggunakan pendekatan transfer learning dengan memanfaatkan model pre-trained ResNet50 yang tersedia di Keras Applications. Model Klasifikasi CNN yang dihasilkan mendapatkan nilai performa akurasi 95%, rata-rata makro precision 95%, rata-rata makro recall 95%, rata-rata makro f1 score 95%. Kata kunci — Citra; Convolutional Neural Network; Klasifikasi; Spesies Ikan; Transfer Learning.