File size: 4,199 Bytes
4583098
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
"""

    Helper functions mainly for data preparation.

"""
import cv2
import numpy as np
from PIL import Image


# Extracting the image and mask from gr.ImageEditor.
def process_gradio_source(editor_data):
    original_image = editor_data["background"].convert("RGB")
    layers = editor_data.get("layers", [])

    full_mask = Image.new("L", original_image.size, 0)
    cropped_region = original_image.copy()

    if not layers:
        return full_mask, cropped_region, original_image

    try:
        layer_image = layers[0]
        layer_pos = (0, 0)

        if layer_image.mode != "RGBA":
            layer_image = layer_image.convert("RGBA")

        alpha_channel = layer_image.split()[-1]

        full_mask = Image.new("L", original_image.size, 0)
        full_mask.paste(alpha_channel, layer_pos)
        full_mask = full_mask.point(lambda p: 255 if p > 128 else 0)

        original_np = np.array(original_image)
        mask_np = np.array(full_mask)

        mask_bool = mask_np > 0
        cropped_array = np.zeros_like(original_np)
        cropped_array[mask_bool] = original_np[mask_bool]

        cropped_region = Image.fromarray(cropped_array)

    except Exception as e:
        print(f"Raise error: {str(e)}")

    return original_image, full_mask, cropped_region


# Get bounding box from the mask.
def get_bbox_from_mask(mask_image):
    mask_array = np.array(mask_image)
    if mask_array.ndim == 3:
        mask_array = cv2.cvtColor(mask_array, cv2.COLOR_RGB2GRAY)
    _, binary_mask = cv2.threshold(mask_array, 127, 255, cv2.THRESH_BINARY)

    contours, _ = cv2.findContours(mask_array, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
    if not contours:
        return None

    largest_contour = max(contours, key=cv2.contourArea)
    x, y, w, h = cv2.boundingRect(largest_contour)
    return (x, y), (x + w, y + h)


# Crop image from the bounding box coord.
def crop_image_from_bb(original_image, top_left, bottom_right):
    x1, y1 = map(int, top_left)
    x2, y2 = map(int, bottom_right)
    if not (0 <= x1 < x2 <= original_image.width and 0 <= y1 < y2 <= original_image.height):
        raise ValueError("Invalid bounding box coordinates")
    crop_box = (x1, y1, x2, y2)
    cropped_image = original_image.crop(crop_box)
    return cropped_image


# Resize image to the target size with zero paddings.
def resize_img_and_pad(input_image, target_size):
    cropped_width, cropped_height = input_image.size
    target_width, target_height = target_size
    scale = min(target_width / cropped_width, target_height / cropped_height)
    new_width = int(cropped_width * scale)
    new_height = int(cropped_height * scale)

    resized_image = input_image.resize((new_width, new_height), Image.BILINEAR)

    padded_image = Image.new("RGB", target_size, (0, 0, 0))

    left_padding = (target_width - new_width) // 2
    top_padding = (target_height - new_height) // 2
    padded_image.paste(resized_image, (left_padding, top_padding))

    return padded_image


def pil_to_np(pil_img):
    if pil_img.mode == 'RGBA':
        rgb = pil_img.convert('RGB')
        alpha = pil_img.split()[3]
        img = np.array(rgb)
        alpha = np.array(alpha)
    else:
        img = np.array(pil_img.convert('RGB'))
    return img


# Helper function considering the VAE compression factor.
def nearest_multiple_of_16(ref_height):
    lower = (ref_height // 16) * 16
    upper = ((ref_height + 15) // 16) * 16
    if ref_height - lower <= upper - ref_height:
        return lower
    else:
        return upper


# Resize to generate the reference context.
def generate_context_reference_image(reference_image_pil, img_width=512):
    reference_image_rgb = pil_to_np(reference_image_pil)
    ref_height, ref_width = reference_image_rgb.shape[0], reference_image_rgb.shape[1]

    ref_height = int((img_width / ref_width) * ref_height)
    ref_height = nearest_multiple_of_16(ref_height)

    reference_context = cv2.resize(reference_image_rgb, (img_width, ref_height))
    reference_new_pil = Image.fromarray(reference_context)
    return reference_new_pil