Impression : Jurnal Teknologi dan Informasi
Vol. 5 No. 2 (2026): Juli 2026

Perancangan Pengenalan Emosi melalui Ekspresi pada Wajah Menggunakan Convolutional Neural Network (CNN)

Khantaqwa Amri (Universitas Pamulang)
Ines Heidiani Ikasari (Universitas Pamulang)



Article Info

Publish Date
17 Oct 2026

Abstract

Pengenalan emosi melalui ekspresi wajah merupakan salah satu bidang penelitian dalam computer vision dan artificial intelligence yang berkembang pesat. Kemampuan sistem dalam mengenali ekspresi wajah manusia dapat diterapkan pada berbagai bidang, seperti pendidikan, pelayanan publik, kesehatan, sistem keamanan, hingga interaksi manusia dengan komputer (Human Computer Interaction). Penelitian ini bertujuan merancang sistem pengenalan emosi manusia secara real-time menggunakan algoritma Convolutional Neural Network (CNN). Dataset yang digunakan terdiri atas tujuh kategori emosi, yaitu angry, disgust, fear, happy, neutral, sad, dan surprise. Proses penelitian meliputi preprocessing dataset, augmentasi data menggunakan ImageDataGenerator, pelatihan model CNN menggunakan TensorFlow Keras, penyimpanan bobot model, serta implementasi deteksi emosi secara real-time menggunakan webcam dan OpenCV. Model CNN dibangun menggunakan beberapa lapisan konvolusi, fungsi aktivasi ReLU, Max Pooling, Dropout, Flatten, Fully Connected Layer, serta Softmax sebagai lapisan klasifikasi. Hasil penelitian menunjukkan bahwa model mampu mengenali ekspresi wajah secara real-time dengan tingkat akurasi yang baik serta mampu mengklasifikasikan tujuh kategori emosi berdasarkan citra wajah yang diperoleh dari kamera. Sistem yang dikembangkan diharapkan dapat menjadi dasar pengembangan aplikasi pengenalan emosi pada berbagai bidang yang membutuhkan interaksi otomatis antara manusia dan komputer.   Facial Emotion Recognition (FER) is an application of computer vision and artificial intelligence that enables a system to automatically recognize facial expressions. This study aims to design and implement a real-time system for recognizing seven emotion categories based on a Convolutional Neural Network (CNN) using a webcam. The dataset consists of seven classes: angry, disgust, fear, happy, neutral, sad, and surprise. Images are processed in grayscale format with a size of 48 × 48 pixels and augmented using ImageDataGenerator. The CNN model is built with convolutional layers, ReLU, Max Pooling, Dropout, Flatten, a fully connected layer, and Softmax, using the Adam optimizer, a batch size of 64, and 50 epochs. Evaluation was based on training accuracy, validation accuracy, training loss, and validation loss. The results show that the model achieved a training accuracy of 95.10% and a validation accuracy of 93.40%, with a training loss of 0.19 and a validation loss of 0.27 at the 50th epoch. The implementation of OpenCV and the Haar Cascade enables real-time face detection and emotion classification via a webcam. Black Box Testing confirmed that all the application’s main functions operate as designed. This study contributes to the field by integrating CNNs with webcams; however, testing on independent datasets as well as measurements of FPS and latency are required in future research.

Copyrights © 2026






Journal Info

Abbrev

jti

Publisher

Subject

Automotive Engineering Chemical Engineering, Chemistry & Bioengineering Civil Engineering, Building, Construction & Architecture Computer Science & IT Electrical & Electronics Engineering

Description

Impression accepts articles in the fields of Electrical Engineering, Mechanical Engineering, Civil Engineering, Marine Technology Industrial Engineering, Marine Fisheries Technology, Agricultural Technology, Informatics Engineering, Information Systems, Computer, Expert systems, Decision Support ...