import cv2
import mediapipe as mp
from mediapipe.tasks import python
from mediapipe.tasks.python import vision
import gradio as gr
import os

eng_ara_mapping ={
'':'not clear',
"ain":'ع',
"aleff":'أ',
"bb":'ب',
"dal":'د',
"dha":'ظ',
"dhad":"ض",
"fa":"ف",
"gaaf":'ق',
"ghain":'غ',
"ha":'ه',
"haa":'ح',
"jeem":'ج',
"kaaf":'ك',
"laam":'ل',
"meem":'م',
"nun":"ن",
"ra":'ر',
"saad":'ص',
"seen":'س',
"sheen":"ش",
"ta":'ت',
"taa":'ط',
"thaa":"ث",
"thal":"ذ",
"waw":'و',
"ya" : "ي",
"zay":'ز',
"khaa":'خ' }

def recognize_gesture(image):
    # Load the gesture recognition model
    model_path = os.path.abspath("arabic_signlanguage_characters_model.task")
    recognizer = vision.GestureRecognizer.create_from_model_path(model_path)

    # Convert image to MediaPipe format
    if isinstance(image, gr.Image):  # Check if image is from Gradio
        image = image.to_ndarray(format="bgr")  # Convert to a NumPy array in BGR format

    # Convert image to MediaPipe format
    image = mp.Image(image_format=mp.ImageFormat.SRGB, data=image)

    # Perform gesture recognition
    recognition_result = recognizer.recognize(image)

    # Extract the top gesture
    top_gesture = recognition_result.gestures[0][0]

    # Return the gesture label and score
    return dict(zip(eng_ara_mapping[top_gesture.category_name], map(float,top_gesture.score)))
                                                            

iface = gr.Interface(
    fn=recognize_gesture,
    inputs=gr.Image(sources=["upload","clipboard"]),  # Input type: image
    outputs=gr.Label(num_top_classes=4),  # Output type: label
    title="Arabic Sign Language Character Recognition",
    description="Upload an image to recognize the character",
)

iface.launch(share=True)  # Launch the interface in a web browser