gradio-app.py

import os
import cv2
import numpy as np
from modules.makeup import Makeup
from PIL import Image

from FaceBoxes import FaceBoxes
import gradio as gr

# Requirement numpy > 1.17

def process_sticker(path_sticker):
    sticker = cv2.imread(path_sticker)
    sticker = cv2.resize(sticker, (256, 256))
    sticker_g = cv2.cvtColor(sticker, cv2.COLOR_BGR2GRAY)
    _, sticker_t = cv2.threshold(sticker_g, 10, 255, cv2.THRESH_BINARY_INV)
    return sticker, sticker_t

def add(img, sticker, sticker_t):
    sticker_no_focus = cv2.add(img, sticker, mask=sticker_t)
    return sticker_no_focus + sticker

def blend(result, ori):
    result = result.astype("uint8")
    result_g = cv2.cvtColor(result, cv2.COLOR_BGR2GRAY)
    _, result_t = cv2.threshold(result_g, 5, 255, cv2.THRESH_BINARY_INV)
    ori_no_focus = cv2.bitwise_and(ori, ori, mask=result_t)
    final = cv2.add(result, ori_no_focus)
    return final

def transform_to_square_bbox(bbox):
    left, top, right, bottom = bbox[:4]
    old_size = (right - left + bottom - top) / 2
    center_x = right - (right - left) / 2.0
    center_y = bottom - (bottom - top) / 2.0 + old_size * 0.03
    size = int(old_size * 1.25)
    roi_box = [0] * 4
    roi_box[0] = center_x - size / 2
    roi_box[1] = center_y - size / 2
    roi_box[2] = roi_box[0] + size
    roi_box[3] = roi_box[1] + size
    return roi_box

def crop(bbox, img):
    x1, y1, x2, y2 = np.uint32(bbox)
    img = img[y1:y2, x1:x2]
    return cv2.resize(img, (256, 256))

def process_input(path_input):
    face_bbox = FaceBoxes()
    img = cv2.imread(path_input.name)
    img_origin = img.copy()
    bbox = face_bbox(img)
    return img, img_origin, bbox

def restore_img(img, img_ori, bbox):
    x1, y1, x2, y2 = np.uint32(bbox)
    size = (x2 - x1, y2 - y1)
    img = cv2.resize(img, size)
    img_ori[y1: y2, x1: x2] = img
    return img_ori

def run(path_input, path_sticker='tests/uv_face_sticker.png'):
    img_A, img_A_origin, bbox = process_input(path_input)
    model = Makeup()
    for bb in bbox:
        bb = transform_to_square_bbox(bb)
        img_A = crop(bb, img_A_origin)
        model.prn_process(img_A)
        A_txt = model.get_texture()
        sticker, sticker_t = process_sticker(path_sticker)
        result = add(A_txt, sticker, sticker_t)
        output = model.render_texture(result)
        final = blend(output, img_A)
        final_of_final = restore_img(final, img_A_origin, bb)
    return final_of_final[:,:,::-1]

inputs = gr.inputs.Image(type='file', label="Original Image")
outputs = gr.outputs.Image(type="file", label="Output Image")

title = "AR Filter"
description = "Make filter like messenger, instagram"

examples = [
            ['tests/g1.jpg'],
            ['tests/g2.jpg'],
            ['tests/g3.jpg'],
]

gr.Interface(run, inputs, outputs, title=title, description=description, examples=examples).launch(debug=False,enable_queue=False, share=True);