[FEATURE] Add comprehensive README.md with setup, execution, metrics, and tech stack

[GabrielLidenor]

Currently, the repository lacks a comprehensive README.md. We need clear documentation explaining how to set up the environment, run the pipeline (training & evaluation), understand the current benchmark performance, and view the tech stack used in this project.

Proposed Structure for README.md

1. Project Overview

• High-level overview of the facial expression recognition pipeline using PyTorch and Hugging Face datasets.

2. Tech Stack & Tools Used

• PyTorch / Torchvision: Model building (ResNet), dataset transforms, training loop, and MPS (Apple Silicon GPU) execution.
• Hugging Face (datasets): Dataset loading and disk caching.
• Scikit-Learn: Evaluation metrics (Accuracy, Precision, Recall, F1, Confusion Matrix).
• Pillow (PIL) & NumPy: Image handling and array transformations.

3. Installation & Setup

• Prerequisites (Python version, virtual environment).
• Commands to clone the repository and install dependencies (pip install -r requirements.txt or poetry install)-

4. How to Run

• Commands for running the complete pipeline (python main.py or equivalent).
• Brief explanation of raw data downloading vs. loading behavior.

5. Model Benchmarks & Metrics

• Summary of current evaluation results (~68% accuracy / F1 score).
• Key metric breakdown (Precision, Recall, F1 macro/weighted) and confusion matrix insights.

Tasks

• Draft README.md covering Setup, Usage, Tech Stack, and Metrics.
• Include code snippets for installation and execution commands.
• Add metric snapshot / confusion matrix performance notes.
• Review and merge into main.