Smart Lock Berbasis Pengenalan Wajah Dan Gestur Tangan Menggunakan LSTM Pada Edge Computing Raspberry Pi 5

Ramadhani, Yuvina Rahma (2026) Smart Lock Berbasis Pengenalan Wajah Dan Gestur Tangan Menggunakan LSTM Pada Edge Computing Raspberry Pi 5. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5024221020_Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (35MB)

Abstract

Perkembangan Internet of Things (IoT) telah mendorong lahirnya teknologi keamanan rumah yang semakin cerdas, adaptif, dan mampu beroperasi secara mandiri. Penelitian ini mengembangkan sistem smart lock berbasis Raspberry Pi 5 yang mengintegrasikan dua metode autentikasi biometrik, yaitu pengenalan wajah dan pengenalan gestur tangan. Sistem dirancang menggunakan pendekatan edge computing sehingga seluruh proses pengolahan citra dan inferensi model dilakukan secara lokal pada perangkat Raspberry Pi tanpa bergantung pada server cloud. Pendekatan ini bertujuan mengurangi latensi, meningkatkan privasi data pengguna, serta memastikan sistem tetap berfungsi meskipun tidak terhubung ke jaringan internet. Kamera digunakan untuk menangkap citra wajah dan urutan gestur tangan yang selanjutnya diproses menggunakan algoritma berbasis Deep Learning. Pengenalan wajah dilakukan menggunakan model deteksi dan klasifikasi berbasis OpenCV, sedangkan pengenalan gestur tangan memanfaatkan model Long Short-Term Memory (LSTM) untuk mengenali pola sekuensial pergerakan tangan. Sistem menghasilkan keputusan akses berupa perintah membuka atau mengunci pintu melalui aktuator yang terhubung dengan Raspberry Pi. Evaluasi dilakukan berdasarkan tiga aspek utama, yaitu akurasi pengenalan wajah dan gestur, ketahanan performa pada berbagai kondisi pencahayaan (lux variation), serta konsumsi daya Raspberry Pi selama beroperasi. Berdasarkan hasil yang diharapkan, sistem mampu memberikan akurasi tinggi dan performa yang stabil pada kedua metode autentikasi meskipun terjadi variasi intensitas cahaya. Dengan kombinasi autentikasi biometrik ganda dan pemrosesan berbasis edge, sistem smart lock ini berpotensi menjadi solusi keamanan rumah yang lebih aman, efisien, dan mudah diimplementasikan pada lingkungan rumah pintar modern.
=================================================================================================================================
The rapid development of the Internet of Things (IoT) has accelerated the emergence of smarter, more adaptive, and autonomous home security technologies. This study presents a Raspberry Pi 5-based smart lock system that integrates two biometric authentication methods: facial recognition and hand gesture recognition. The system is designed using an edge computing architecture, enabling all image processing and model inference tasks to be executed locally on the Raspberry Pi without relying on external cloud servers. This approach reduces latency, enhances user data privacy, and ensures continuous operation even in the absence of an internet connection. A camera captures facial images and hand gesture sequences, which are processed using Deep Learning-based algorithms. Facial recognition is implemented through OpenCV-based detection and classification models, while hand gesture recognition employs a Long Short-Term Memory (LSTM) network to identify sequential movement patterns. The system determines access authorization by issuing lock or unlock commands to an actuator connected to the Raspberry Pi. Performance evaluation focuses on three key aspects: the accuracy of facial and gesture recognition, system robustness under varying lighting conditions (lux variation), and the Raspberry Pi's power consumption during continuous operation. Based on the expected outcomes, the proposed system is anticipated to achieve high recognition accuracy and stable performance for both authentication methods despite changes in illumination. By combining dual biometric authentication with edge-based processing, the proposed smart lock offers a secure, efficient, and practical solution for modern smart home applications.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Smart Lock, Raspberry Pi 5, Pengenalan Wajah, Gestur Tangan, LSTM., Smart Lock, Raspberry Pi 5, Face Recognition, Hand Gesture Recognition, LSTM.
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. Q Science > QA Mathematics > QA76.6 Computer programming. Q Science > QA Mathematics > QA76.758 Software engineering T Technology > T Technology (General)
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Computer Engineering > 90243-(S1) Undergraduate Thesis
Depositing User:	Yuvina Rahma Ramadhani
Date Deposited:	17 Jul 2026 06:12
Last Modified:	17 Jul 2026 06:12
URI:	http://repository.its.ac.id/id/eprint/134920

Actions (login required)

View Item