Improved Accuracy using CNN (Issue: #860)

[PriyanshuLathi]

Related Issues or Bug

• The current face expression recommendation system uses MobileNet, which has an accuracy of 55%. There is a need to evaluate a custom CNN model built from scratch, which achieves an accuracy of 65.33%. Additionally, confusion matrix and ROC plots for model evaluation are missing.

Fixes: #860

---

Proposed Changes

• Added a custom CNN model for face expression recommendation, which achieves an accuracy of 65.33%.
• Integrated performance evaluation using AUC plots for both MobileNet and the custom CNN model.
• Added confusion matrix and ROC plots to compare the performance of both models.

---

Additional Info

• The custom CNN model is designed as an alternative to MobileNet. AUC plots, confusion matrix, and ROC scores are now included to evaluate and compare both models' performances.

---

Screenshots

Original (MobileNet)	Updated (Custom CNN)
Din't have any classification report

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated (UTC)
ml-nexus	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Nov 10, 2024 1:15pm

[github-actions]

👋 Thank you for opening this pull request! We appreciate your contribution to improving this project. Your PR is under review, and we'll get back to you shortly.
Don't forget to mention the issue you solved!