Facial Expression-Based Emotion Detection: A CNN-Based Approach for Real-Time Emotion Recognition

The result for this study proves that CNN based DL models are effective for facial expression recognition, achieving reasonable accuracy in classifying eight emotion categories. The systems real time performance demonstrates its potential for practical applications in human-computer interaction. However, several limitations were identified: Dataset Constraints: The FER-2013 dataset, while widely used, as limitation in terms of diversity and real world representation. The grayscale nature of images and the controlled conditions under collected may not fully capture the complexity of real-world facial expressions. Class Imbalance: The dataset exhibit class imbalance, with some emotions having significantly more samples than others. This can lead to biased predictions favoring the majority classes. Environmental Factors: Real-world deployment introduces challenges such as differential lighting conditions, head poses, and occlusions that were not fully represented in the training data. Model Architecture: While the CNN architecture used in this research is effective, more advanced architectures like ResNet, EfficientNet, or hybrid CNN-RNN models could potentially achieve higher accuracy

1.Krishna Kant, Dipti B. Shah Lightweight Attention Based Deep CNN Framework for Human Facial Emotion Detection from Video Sequences,Received: 23 July 2024 / Accepted: 14 November 2024© The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd. 2024. 2.    S. Woo, J. Park, J. Lee, and I. S. Kweon, ‘‘CBAM: Convolutional block attention module,’’ in Proc. Eur. Conf. Comput. Vis. (ECCV), Jan. 2018, pp. 3–19. 3.   D. Yang, Z. Chen, Y. Wang, S. Wang, M. Li, S. Liu, X. Zhao, S. Huang, Z. Dong, P. Zhai, and L. Zhang, ‘‘Context de-confounded emotion recognition,’’ in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit. (CVPR), Jun. 2023, pp. 19005–19015. 4.   Z.Yang and S. S. Narayanan, ‘‘Modeling dynamics of expressive body gesture in dyadic interactions,’’ IEEE Trans Affect. Comput., vol.8, no.3, pp. 369–381, Jul. 2017. 5.   P. B a r r o s , D . J i r a k , C . We b e r , a n d S . We r m t e r , ‘‘Multimodalemotionalstate recognition using sequence-dependent deep hierarchical features,’’ Neural Netw., vol.72, pp. 140–151, Dec. 2015.. 6.  L. F. Barrett, B. Mesquita, and M. Gendron, ‘‘Context in emotion perception,’’ Curr. Dir. Psychol. Sci., vol. 20, no.5, pp. 286–290, Oct. 2011. 7.  S. Zhang, Y. Pan, and J. Z. Wang, ‘‘Learning emotion representations from verbal and nonverbal communication,’’ in Proc.IEEE/CVF Conf.comput. Vis.Pattern Recognit. (CVPR), Jun. 2023, pp. 18993–19004. 8.   F. Chollet, ‘‘Xception: DL along with depthwise separable convolutions,’’ in Proc. IEEE conf. comput. Vis. pattern recognit. (CVPR), Jul.2017, pp. 1800–1807. 9.   S. Woo, J. Park, J. Lee, and I. S. Kweon, ‘‘CBAM: Convolutional block attention module,’’ in Proc.Eur. Conf.Comput. Vis. (ECCV), Jan. 2018, pp. 3–19. 10.  K. Xu et al., ‘‘Show, attend and tell: Neurl image caption generation with visual attentio n,’’ in Proc. Int. Conf. Mach. Learn., 2015, pp. 2048–