Hugging Face's logo

Spaces:
ImAMJayKIM
/
Mini-ImageNet
Sleeping

App Files Community
Mini-ImageNet
File size: 2,361 Bytes
c1596ac
 
 
 
 
 
 
 
 
 
 
 
 
2729151
c1596ac
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import os
import cv2
import torch
import torch.nn as nn
import numpy as np

from PIL import Image

from pytorch_grad_cam import GradCAM
from pytorch_grad_cam.utils.image import (
    show_cam_on_image
)

from src.transforms.image_transform import (
    get_classification_valid_transform
)




class SwinClassifierWrapper(nn.Module):

    def __init__(self, model):
        super().__init__()
        self.model = model

    def forward(self, images):

        features = self.model.backbone(images)

        features = features.view(
            features.size(0),
            -1
        )

        logits = self.model.classifier(features)

        return logits




def reshape_transform(tensor):

    # Swin-T feature output: B, H, W, C
    # Grad-CAM expects: B, C, H, W
    if tensor.ndim == 4:

        tensor = tensor.permute(
            0,
            3,
            1,
            2
        )

    return tensor




def save_gradcam(

    model,

    image_path,

    save_path,

    device

):

    model.eval()

    for param in model.backbone.parameters():
        param.requires_grad = True

    for param in model.classifier.parameters():
        param.requires_grad = True

    gradcam_model = SwinClassifierWrapper(
        model
    ).to(device)

    gradcam_model.eval()

    transform = (
        get_classification_valid_transform()
    )

    image = Image.open(
        image_path
    ).convert("RGB")

    image = image.resize(
        (224, 224)
    )

    image_np = (
        np.array(image)
        .astype(np.float32)
        / 255.0
    )

    tensor = transform(
        image
    ).unsqueeze(0).to(device)

    target_layer = (
        model.backbone.features[-1][-1].norm2
    )

    cam = GradCAM(
        model=gradcam_model,
        target_layers=[target_layer],
        reshape_transform=reshape_transform
    )

    grayscale_cam = cam(
        input_tensor=tensor
    )[0]

    visualization = show_cam_on_image(
        image_np,
        grayscale_cam,
        use_rgb=True
    )

    os.makedirs(
        os.path.dirname(save_path),
        exist_ok=True
    )

    cv2.imwrite(
        save_path,
        cv2.cvtColor(
            visualization,
            cv2.COLOR_RGB2BGR
        )
    )