import cv2
import gradio as gr
from cvzone.HandTrackingModule import HandDetector
from cvzone.ClassificationModule import Classifier
import numpy as np
import math

# Load HandDetector and Classifier
detector = HandDetector(maxHands=1)
classifier = Classifier("keras_model.h5", "labels.txt")
offset = 20
imgSize = 300

# Try different camera indices until a valid one is found
camera_index = 0
cap = None

while cap is None or not cap.isOpened():
    cap = cv2.VideoCapture(camera_index)
    if cap is None or not cap.isOpened():
        camera_index += 1
        print(f"Error: Could not open camera with index {camera_index - 1}. Trying index {camera_index}.")

if cap.isOpened():
    print(f"Camera opened successfully with index {camera_index}.")
else:
    print("Error: No valid camera index found.")
    exit()

def classify_hand(img):
    imgOutput = img.copy()

    # Detect hands
    hands, _ = detector.findHands(img)

    if hands:
        hand = hands[0]
        x, y, w, h = hand['bbox']

        imgWhite = np.ones((imgSize, imgSize, 3), np.uint8) * 255
        imgCrop = img[y - offset:y + h + offset, x - offset:x + w + offset]

        if imgCrop.size != 0:
            imgCropShape = imgCrop.shape
            aspectRatio = h / w

            if aspectRatio > 1:
                k = imgSize / h
                wCal = math.ceil(k * w)
                imgResize = cv2.resize(imgCrop, (wCal, imgSize))
                imgResizeShape = imgResize.shape
                wGap = math.ceil((imgSize - wCal) / 2)
                imgWhite[:, wGap: wCal + wGap] = imgResize

            else:
                k = imgSize / w
                hCal = math.ceil(k * h)
                imgResize = cv2.resize(imgCrop, (imgSize, hCal))
                imgResizeShape = imgResize.shape
                hGap = math.ceil((imgSize - hCal) / 2)
                imgWhite[hGap: hCal + hGap, :] = imgResize

            # Get hand gesture prediction
            prediction, index = classifier.getPrediction(imgWhite, draw=False)

            # Draw bounding box and label
            cv2.rectangle(imgOutput, (x - offset, y - offset - 70), (x - offset + 400, y - offset + 60 - 50), (0, 255, 0),
                          cv2.FILLED)
            cv2.putText(imgOutput, labels[index], (x, y - 30), cv2.FONT_HERSHEY_COMPLEX, 2, (0, 0, 0), 2)
            cv2.rectangle(imgOutput, (x - offset, y - offset), (x + w + offset, y + h + offset), (0, 255, 0), 4)

    return imgOutput

iface = gr.Interface(fn=classify_hand, inputs='webcam', outputs='image', live=True)
iface.launch()