File size: 6,818 Bytes
cfbb05a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
import numpy as np
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split
from PIL import Image
import pickle
import os
import tqdm
from collections import Counter
import random
import io

from lime import lime_image
from skimage.segmentation import mark_boundaries
import shap
from typing import List, Tuple
import torch


class Utils:
    def __init__(self, in_dir: str):
        self.in_dir = in_dir

    @staticmethod
    def lime_explain_instance(model, image, num_samples: int, num_features: int):
        """
        Generate LIME explanation for a given image.
        :param model: Model to explain.
        :param image: Input image as a NumPy array.
        :param num_samples: Number of samples for LIME explanation.
        :param num_features: Number of features to highlight in explanation.
        :return: Marked boundaries image showing explanation.
        """
        explainer = lime_image.LimeImageExplainer()
        
        def predict_fn(images):
            model.eval()
            with torch.no_grad():
                images = torch.tensor(images).permute(0, 3, 1, 2).float()
                outputs = model(images)
                return outputs.numpy()
        
        explanation = explainer.explain_instance(
            image.astype("double"),
            predict_fn,
            top_labels=3,
            hide_color=0,
            num_samples=num_samples,
        )
        temp, mask = explanation.get_image_and_mask(
            explanation.top_labels[0],
            positive_only=False,
            num_features=num_features,
            hide_rest=False,
        )
        return mark_boundaries(temp / 2 + 0.5, mask)

    @staticmethod
    def shapley_explain_instance(
        model, image, labels: List[str] = None, evals: int = 5000, top_labels: int = 3
    ):
        """
        Generate SHAP explanation for a given image.
        :param model: Model to explain.
        :param image: Input image as a NumPy array.
        :param labels: List of class labels.
        :param evals: Number of evaluations for SHAP.
        :param top_labels: Number of top labels to explain.
        :return: SHAP explanation as an image.
        """
        masker = shap.maskers.Image("inpaint_ns", image[0].shape)
        explainer = shap.Explainer(model, masker, output_names=labels)
        shap_values = explainer(
            image,
            max_evals=evals,
            batch_size=100,
            outputs=shap.Explanation.argsort.flip[:top_labels],
        )

        plt.figure()
        shap.image_plot(shap_values, show=False)
        buf = io.BytesIO()
        plt.savefig(buf, format="png")
        buf.seek(0)
        img = Image.open(buf)
        buf.close()
        return np.array(img)


class Processing:
    @staticmethod
    def load_data(file_path: str) -> np.ndarray:
        """Load data from a file."""
        if not os.path.exists(file_path):
            raise FileNotFoundError(f"File not found: {file_path}")
        return np.load(file_path)

    @staticmethod
    def save_to_pickle(data: List, file_path: str):
        """Save data to a pickle file."""
        with open(file_path, "wb") as f:
            pickle.dump(data, f)

    @staticmethod
    def load_from_pickle(file_path: str):
        """Load data from a pickle file."""
        if not os.path.exists(file_path):
            raise FileNotFoundError(f"File not found: {file_path}")
        with open(file_path, "rb") as f:
            return pickle.load(f)

    @staticmethod
    def norm_image(x: np.ndarray) -> np.ndarray:
        """Normalize an image to the range [0, 1]."""
        return x / 255.0

    @staticmethod
    def denorm_image(x: np.ndarray) -> np.ndarray:
        """Denormalize an image to the range [0, 255]."""
        return x * 255.0

    @staticmethod
    def to_categorical(y: np.ndarray, num_classes: int) -> np.ndarray:
        """Convert an array of integers to a one-hot encoded array."""
        return np.eye(num_classes)[y]

    @staticmethod
    def from_categorical(y: np.ndarray) -> np.ndarray:
        """Convert a one-hot encoded array to an array of integers."""
        return np.argmax(y, axis=1)

    @staticmethod
    def zip_images(directory: str, label: int) -> List[Tuple[np.ndarray, int]]:
        """Read images from a directory and return as a list of (image, label) tuples."""
        data = []
        for file_name in tqdm.tqdm(
            os.listdir(directory), desc=f"Processing {directory}"
        ):
            img_path = os.path.join(directory, file_name)
            try:
                img = Image.open(img_path).convert("RGB")
                data.append((np.array(img), label))
            except Exception as e:
                print(f"Error loading image {img_path}: {e}")
        return data


class Visualization:
    @staticmethod
    def plot_distribution(train_data, val_data, test_data, class_names):
        """Plot distribution of data after splitting into train, validation, and test sets."""
        train_labels = [label for _, label in train_data]
        val_labels = [label for _, label in val_data]
        test_labels = [label for _, label in test_data]

        train_counter = Counter(train_labels)
        val_counter = Counter(val_labels)
        test_counter = Counter(test_labels)

        labels = sorted(set(train_labels + val_labels + test_labels))
        train_counts = [train_counter[label] for label in labels]
        val_counts = [val_counter[label] for label in labels]
        test_counts = [test_counter[label] for label in labels]

        x = range(len(labels))
        width = 0.25

        plt.figure(figsize=(10, 6))
        plt.bar(x, train_counts, width=width, label="Train", color="blue")
        plt.bar(
            [p + width for p in x],
            val_counts,
            width=width,
            label="Validation",
            color="orange",
        )
        plt.bar(
            [p + width * 2 for p in x],
            test_counts,
            width=width,
            label="Test",
            color="green",
        )

        plt.xlabel("Classes")
        plt.ylabel("Number of Samples")
        plt.title("Distribution of Data After Split")
        plt.xticks([p + width for p in x], [class_names[label] for label in labels])
        plt.legend()
        plt.show()

    @staticmethod
    def plot_n_images(images: np.ndarray, n: int, img_per_row: int, save_path: str):
        """Plot n images with a specified number of images per row."""
        rows = (n + img_per_row - 1) // img_per_row
        plt.figure(figsize=(img_per_row * 2, rows * 2))
        for i in range(n):
            plt.subplot(rows, img_per_row, i + 1)
            plt.imshow(images[i])
            plt.axis("off")
        plt.savefig(save_path, bbox_inches="tight")