Emotion-Classification

---

Problem Statement

The problem statement is to predict face emotion of a human being using Haarcascade algorithm and CNN based trained models.

In this project, we will analyse various face images that have been collected and classify them into following categories.

• Angry
• Disgusted
• Fearful
• Happy
• Neutral
• Sad
• Surprised

---

Business Problem Overview

In many applications, it is a must to understand behavior of a person using facial expression. Using this facial classification, it is possible to understand customers review without having him / her to wite / jot down current mood. Out of the many applications, this can be also used to understand a person behavior while working on a desk to understand whether employee is satisfied during an interview, call or a meeting.

Taking this business problem in mind, we will move ahead in understanding this project in even more depth.

---

Project Pipeline

For this we are using 2 models.

1. Face Detection Using Haar-cascade Algorithm
   • This module includes detecting as many faces inside the image
   • Creating bounding box around these faces
2. CNN Trained Model:
   • Library Management
   • Sourcing Data
   • Model Training (With Hyperparameter Tuning)
3. Using Trained Model for detecting Face(s) & Emotion(s) on WebCam

Note:

• User can run main.py to detect emotions.
• Webcam is needed to capture live stream

---

Conclusion

Accuracy of Trained Models Till Date:

Date	Train Accuracy (%)	Validation Accuracy (%)
18 April 2023	95.69	62.91
19 April 2023	---	---