app.py

import cv2
import imutils
import gradio as gr
import numpy as np


face_detector = cv2.CascadeClassifier(cv2.data.haarcascades + "haarcascade_frontalface_default.xml")

def detect_faces(img, size, neighbours, scale):
    frame = np.array(img) 
    frame = frame[:, :, ::-1].copy()
    frame = imutils.resize(frame, width=500)
    gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)

    faceRects = face_detector.detectMultiScale(
        gray, scaleFactor=scale, minNeighbors=int(neighbours), minSize=(int(size), int(size))
    )

    box_data = []
    class_labels = {0: "face"}

    for (x, y, w, h) in faceRects:
        frame = cv2.rectangle(frame, (x, y), (x+w, y+h), (255, 0, 0), 2)
        midX = int(x + w / 2)
        midY = int(y + h / 2)
        box = {
            "position": {"middle": [midX, midY], "width": float(w), "height": float(h)},
            "domain": "pixel",
            "class_id": 0
        }
        box_data.append(box)

    re_im = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
    return re_im

image = gr.components.Image()
out_im = gr.components.Image()

size_slider = gr.components.Slider(minimum=5, maximum=50, value=30, step=5, label="MinSize in Pixel")
neighbour_slider = gr.components.Slider(minimum=1, maximum=20, value=5, step=1, label="Min Number of Neighbours")
scale_slider = gr.components.Slider(minimum=1.1, maximum=2.0, value=1.3, step=0.1, label="Scale Factor")

Iface = gr.Interface(
    fn=detect_faces,
    inputs=[image, size_slider, neighbour_slider, scale_slider],
    outputs=out_im,
    title="Haar Cascade Object Detection",
    description="Face Detection with Haar Cascades using OpenCV"
).launch()