Assignment.py

"""
📸 Real-Time Object Detection with HAAR Cascades (Face, Eyes, Smile, and More)

================================================================================

This script demonstrates real-time object detection using OpenCV's HAAR cascade classifiers.
It highlights how to use pre-trained models to detect multiple types of visual features
such as faces, eyes, and smiles, and can be extended to detect other objects like:
    - Full human body
    - Upper body
    - Cars
    - License plates
    - Cats, dogs, etc.

The program captures video from the webcam, detects objects in each frame, and overlays
bounding boxes and labels to indicate the detection results.

--------------------------------------------------------------------------------

🎯 Detection Capabilities:
--------------------------
✅ Face Detection:
    - Detects human faces using `haarcascade_frontalface_default.xml`
    - Draws a green rectangle around each face
    - Optionally labels the face

✅ Eye Detection:
    - Detects eyes within the face region using `haarcascade_eye.xml`
    - Draws blue rectangles around eyes
    - Can be used for blink or gaze detection extensions

✅ Smile Detection:
    - Detects smiling within the face region using `haarcascade_smile.xml`
    - Draws yellow rectangles around the smile
    - Can be used to trigger events like auto-capture

✅ Extendable Object Detection:
    - Can include other HAAR cascades, such as:
        - `haarcascade_fullbody.xml`
        - `haarcascade_upperbody.xml`
        - `haarcascade_car.xml`
        - `haarcascade_russian_plate_number.xml`
        - `haarcascade_frontalcatface.xml`

    - Simply load the cascade and apply `detectMultiScale()` as shown in the script.

--------------------------------------------------------------------------------

📦 Requirements:
---------------
- Python 3.x
- OpenCV (`cv2`)

Install OpenCV:
    pip install opencv-python

--------------------------------------------------------------------------------

🧠 Core Concepts Used:
----------------------
- Grayscale conversion: for faster processing
- `cv2.CascadeClassifier`: for loading XML-based pre-trained models
- `detectMultiScale()`: for object detection at different scales
- Region of Interest (ROI): to restrict detection to specific areas
- Drawing and labeling: with `cv2.rectangle()` and `cv2.putText()`

--------------------------------------------------------------------------------

🛠️ Usage Instructions:
----------------------
1. Ensure the HAAR XML files are correctly placed in your project.
2. Run the script.
3. A window will display the webcam feed with live detections.
4. Press the **'q' key** to quit the program.

--------------------------------------------------------------------------------

💡 Notes:
--------
- Detection accuracy depends on:
    - Lighting conditions
    - Camera quality
    - Cascade parameters (`scaleFactor`, `minNeighbors`)
- For better results in production apps, consider using:
    - DNN models (e.g., OpenCV DNN, MediaPipe, or YOLO)
    - Tracking algorithms for object persistence

--------------------------------------------------------------------------------

👨‍💻 Author:
-----------
- Developed as part of an OpenCV exploration project.
- Extend, reuse, or modify freely for learning or custom applications.

"""

import cv2
import tkinter as tk
from tkinter import messagebox

# Path to your HAAR cascades
CASCADE_DIR = "Phase_7/HAAR_CASCADES/"

CASCADE_FILES = {
    "Face": "haarcascade_frontalface_default.xml",
    "Eyes": "haarcascade_eye.xml",
    "Smile": "haarcascade_smile.xml",
    "Full Body": "haarcascade_fullbody.xml",
    "Upper Body": "haarcascade_upperbody.xml",
    "Car": "haarcascade_car.xml",
    "License Plate": "haarcascade_russian_plate_number.xml",
    "Cat Face": "haarcascade_frontalcatface.xml",
}


class HaarApp:
    def __init__(self, root):
        self.root = root
        self.root.title("HAAR Cascade Detection")
        self.root.geometry("400x400")

        tk.Label(root, text="Select objects to detect:", font=("Arial", 14)).pack(
            pady=10
        )

        self.vars = {}
        for name in CASCADE_FILES:
            var = tk.BooleanVar(value=False)
            cb = tk.Checkbutton(root, text=name, variable=var, font=("Arial", 12))
            cb.pack(anchor="w", padx=20)
            self.vars[name] = var

        tk.Button(
            root,
            text="Start Detection",
            command=self.start_detection,
            font=("Arial", 14),
            bg="green",
            fg="white",
        ).pack(pady=20)

    def start_detection(self):
        selected = [name for name, var in self.vars.items() if var.get()]
        if not selected:
            messagebox.showwarning(
                "No Selection", "Please select at least one detector!"
            )
            return

        cascades = {}
        for name in selected:
            path = CASCADE_DIR + CASCADE_FILES[name]
            cascade = cv2.CascadeClassifier(path)
            if cascade.empty():
                messagebox.showerror(
                    "Error", f"Could not load cascade: {CASCADE_FILES[name]}"
                )
                return
            cascades[name] = cascade

        cap = cv2.VideoCapture(0)
        if not cap.isOpened():
            messagebox.showerror("Error", "Could not open webcam.")
            return

        while True:
            ret, frame = cap.read()
            if not ret:
                break
            gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)

            for name, cascade in cascades.items():
                # tune parameters depending on cascade type
                scale = 1.3
                neighbors = 5
                min_size = (30, 30)

                if name == "Smile":
                    scale, neighbors = 1.7, 22
                elif name == "Eyes":
                    scale, neighbors = 1.1, 10

                rects = cascade.detectMultiScale(
                    gray, scale, neighbors, minSize=min_size
                )
                color = (0, 255, 0) if name == "Face" else (255, 0, 0)
                for x, y, w, h in rects:
                    cv2.rectangle(frame, (x, y), (x + w, y + h), color, 2)
                    cv2.putText(
                        frame, name, (x, y - 5), cv2.FONT_HERSHEY_SIMPLEX, 0.6, color, 2
                    )

            cv2.imshow("HAAR Detection", frame)

            if cv2.waitKey(1) & 0xFF == ord("q"):
                break

        cap.release()
        cv2.destroyAllWindows()


if __name__ == "__main__":
    root = tk.Tk()
    app = HaarApp(root)
    root.mainloop()