Add shape classification using Zernike Moments + Streamlit demo

[Mandar808]

🔄 Pull Request: Shape Classification using Zernike Moments with NLP Extension

This PR introduces a comprehensive example for shape classification using Zernike Moments on a custom image dataset (circles, squares, triangles), with an additional NLP-based interaction mode.

---

✅ Key Contributions:

🧠 Core Features:

• examples/shape_classifier.py
  → Trains an SVM classifier using Zernike Moments extracted from shape images.
  → Includes prediction confidence metrics and visualizations.

• shape_app.py
  → A Streamlit web app that allows users to classify shapes by uploading an image.
  → Enhanced to accept natural language text input and classify shapes based on textual descriptions (e.g., "three sides", "perfectly round").

• text_to_shape_classifier.joblib
  → An NLP-based classifier trained to map textual shape descriptions to shape labels, enabling a novel text-to-shape prediction capability.

• Visual Confidence Tools
  → Added confidence_plot.png and confidence_histogram.png to visualize prediction certainty per sample and per class.

• datasets/shapes/
  → Sample dataset of binary shape images used for training.

• Trained Models
  → zernike_shape_svm_model.joblib for image-based classification
  → label_encoder.joblib for label decoding
  → text_to_shape_classifier.joblib for NLP-driven shape understanding

---

🧠 🔤 NLP Integration:

This project now bridges vision and language by allowing users to describe a shape using plain English, which is classified using an NLP pipeline.
This represents a unique cross-domain enhancement—using natural language understanding (NLU) to trigger shape classification in a computer vision pipeline.

---

📚 README:

Instructions updated with:

• How to run the shape classifier via image or text input
• Model usage and architecture
• Setup steps for running the Streamlit app

---

🚀 Summary:

This PR enhances the pyfeats library by:

• Providing an end-to-end demo of Zernike moment-based shape recognition
• Introducing a novel text-to-vision pipeline via NLP
• Making the solution easily explorable through a Streamlit app