سامانه همایش ها

English

صفحه اصلی / سی و دومین کنفرانس ملی و دهمین کنفرانس بین المللی مهندسی زیست پزشکی ایران

A Real-Time Integrated Framework for Face Detection, Gender, and Emotion Recognition Using Convolutional Neural Networks

نویسندگان :

Mostafa Asgarinejad¹ Elias Ebrahimzadeh² Vida Mirabolfathi³ Lila Rajabion⁴ Hamid Soltanian-Zadeh⁵

1- موسسه آموزش عالی علوم شناختی 2- دانشگاه تهران دانشکده علوم مهندسی 3- موسسه آموزش عالی علوم شناختی 4- دانشگاه نیویورک آمریکا 5- دانشگاه تهران دانشکده علوم مهندسی

کلمات کلیدی :

Convolutional neural networks،Facial expressions،Gender Classification،Facial Emotion Recognition،Deep Learning

چکیده :

This paper presents a real-time convolutional neural network (CNN) framework for simultaneous facial emotion and gender recognition. The proposed architecture is designed to process facial images and classify them into six basic emotion categories (happiness, sadness, anger, fear, surprise, disgust) and two gender classes (male, female). The model employs a series of convolutional, pooling, and fully connected layers to hierarchically extract and classify discriminative facial features. When evaluated on publicly available benchmarks, the system achieves state-of-the-art performance, reporting 85% accuracy on the FER-2013 emotion recognition dataset. To enhance interpretability, a guided backpropagation visualization technique is integrated, enabling real-time analysis of learned features and weight dynamics across layers. We demonstrate that combining modern CNN design, advanced regularization, and visual explainability is essential for bridging the gap between offline performance and real-time deployment. The framework is validated through implementation in an integrated vision system capable of unified face detection, gender classification, and emotion recognition in a single forward pass. With applications in social robotics, human-computer interaction, and affective computing, this work provides both a scalable CNN architecture and an open-source implementation to support further innovation in real-time affective vision systems.