Scribd
Upload
61 views16 pages

Project Presentation

Uploaded by

Angrejji65
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
61 views16 pages

Project Presentation

Uploaded by

Angrejji65
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 16

Chandigarh

University

REAL-TIME FACIAL
EMOTION RECOGNITION

Muskan Shangari-21BCS11144
Disket Angmo-21BCS11243
Saurav-21BCS7660
Pushpendra Sharma-21BCS11206
Overview
 Introduction  Literature review
 Problem  Results
 Objectives  Conclusion
 Methodology  Future work
 Implementation 
References
Introduction
This project focuses on abstract facial emotion recognition using Convolutional
Neural Networks (CNNs). By leveraging deep learning techniques, the goal is to
develop a model capable of detecting and classifying emotions expressed through
facial expressions. The proposed approach aims to harness the power of CNNs in
capturing relevant features and achieving accurate emotion recognition.
Proble
ms
The lack of robustness and generalization in existing facial
emotion recognition systems poses a challenge. The proposed
project aims to address this issue by developing a CNN-based
approach that can effectively recognize and classify abstract
emotions expressed through facial expressions. Limited
availability of diverse and well-balanced datasets for abstract
facial emotion recognition hinders the development and
evaluation of accurate models. This project seeks to overcome
this challenge by either curating or augmenting existing
datasets to ensure a comprehensive representation of various
emotions, demographics, and facial expressions.
Objectives

• Develop a CNN-based model for abstract facial emotion recognition with


improved robustness and generalization.
• Enhance the accuracy and performance of facial emotion recognition by
leveraging deep learning techniques and advanced network architectures.
• Curate or augment existing datasets to create a diverse and balanced collection of
labeled facial images representing a wide range of abstract emotions.
Literature review

The literature review reveals that CNN-based approaches have shown promising results in
facial emotion recognition tasks, highlighting the importance of feature extraction, dataset
diversity, and model optimization techniques in achieving accurate and robust emotion
classification.
Ex-
"Real-time emotion recognition using multimodal fusion of facial expressions and
speech" by Tao et al. (2020): This work proposed a multimodal fusion approach to real-time
emotion recognition by combining facial expressions and speech signals.
Methodology
1) Preprocessing the facial images by resizing and normalizing them for
input to the CNN model.
2) Designing a deep learning architecture with convolutional layers to
extract relevant features from the facial images.
3) Training the CNN model using a large dataset of labeled facial
expressions and optimizing it through back propagation and gradient
descent.
4) Evaluating the trained model's performance on a separate test set to
detect the expressions.
Dataset
0 The dataset for this project will consist of diverse
facial images expressing various abstract
1 emotions across different demographics.
Method
0 The method comprises preprocessing facial images,

2 designing a CNN architecture, training the model on


labeled data, evaluating performance, and fine-
tuning for accurate and robust abstract facial
emotion recognition.
Results
Conclusion
In conclusion, the use of Convolutional Neural Networks
(CNNs) for abstract facial emotion recognition shows
promise, but further research is needed to improve
accuracy, interpretability, and real-world applicability in
diverse contexts. This project tested the performance of
CNNs for sound classification on the FER 2013 dataset.
The dataset had class imbalance and variations in audio
file properties. Four CNN models with different activation
functions were implemented
Future works
1 Explore fine-grained emotion recognition beyond basic categories.
2 Develop explainable AI methods for transparent predictions.
Investigate long-term emotion analysis considering temporal dynamics.
3 Integrate contextual information for context-aware emotion recognition.
References

 Ekman, P., Friesen, W. V. (1971). Faces and emotions are the same across
cultures. Journal of Personality and Social Psychology, 17(2), 124-129.

 Simonyan, K. & Zisserman, A. (2014). Very deep convolutional networks


for large-scale image recognition. arXiv preprint arXiv:1409.1556.

 Zhang, Z., Song, Y., Qi, H. (2020). Advances in deep learning for facial
expression recognition: a comprehensive survey. IEEE Transactions on
Affective Computing, 11(4), 705- 727.

 Liu, X., Wang, Yi, and Gong, Y. (2021). Deep emotional cognition: An
investigation. ACM Computing Survey (CSUR), 54(2), 1-38.
Thank you.

You might also like