app.py

import gradio as gr
import numpy as np
from PIL import Image
import cv2
import time
import skops.io as sio
def single_image(image,mode):
    img = image.resize((128, 128))
    if mode=="rgb":
        return np.array(img).flatten()
    elif mode=="hsv":
        return np.array(img.convert('HSV')).flatten()
    else:
        kmeans = sio.load('Models/kmeans.skops')
        num_clusters=kmeans.n_clusters
        sift = cv2.SIFT_create()
        img = np.array(img.convert("L"))
        keypoints, descriptors = sift.detectAndCompute(img, None)
        words = kmeans.predict(descriptors)
        hist, _ = np.histogram(words, bins=num_clusters, range=(0, num_clusters))
        return np.array(hist)
def Classify(img, pre,model):
    start_time = time.time() 
    image = Image.open(img)
    file="Models/"+model+'_'+pre+'.skops'
    loaded_model = sio.load(file)
    predictions = loaded_model.predict(single_image(image,pre).reshape(1,-1))[0]
    end_time = time.time()  
    elapsed_time_microseconds = (end_time - start_time) * 1_000  

    return  predictions,(end_time - start_time) * 1_000


interface = gr.Interface(
    fn=Classify,  
    inputs=[
       gr.Image(type="filepath"),
        gr.Radio(
            ["rgb", "hsv", "sift"],
            label="Preprocessing",
            info="Choose one" 
        )  , gr.Radio(
            ["dt", "rf", "gb"],
            label="ML Model (Dicision Tree, Random Forest, Gradient Boosting)",
            info="Choose one"  
        ) 
    ],
    outputs=[
            gr.Textbox(label="Class"),
          gr.Textbox(label="Time (milliseconds)")

    ]
)

# Launch the interface
interface.launch()