Image Classification
timm
File size: 5,591 Bytes
eb8bf23
 
 
938a123
 
 
 
eb8bf23
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
938a123
 
 
80578f6
938a123
eb8bf23
 
 
 
938a123
 
 
eb8bf23
 
938a123
 
 
2c1eb35
938a123
5cb75da
eb8bf23
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import json

import gradio as gr
from PIL import Image
import safetensors.torch
import timm
from timm.models import VisionTransformer
import torch
from torchvision.transforms import transforms
from torchvision.transforms import InterpolationMode
import torchvision.transforms.functional as TF

torch.set_grad_enabled(False)

class Fit(torch.nn.Module):
    def __init__(
        self,
        bounds: tuple[int, int] | int,
        interpolation = InterpolationMode.LANCZOS,
        grow: bool = True,
        pad: float | None = None
    ):
        super().__init__()

        self.bounds = (bounds, bounds) if isinstance(bounds, int) else bounds
        self.interpolation = interpolation
        self.grow = grow
        self.pad = pad

    def forward(self, img: Image) -> Image:
        wimg, himg = img.size
        hbound, wbound = self.bounds

        hscale = hbound / himg
        wscale = wbound / wimg

        if not self.grow:
            hscale = min(hscale, 1.0)
            wscale = min(wscale, 1.0)

        scale = min(hscale, wscale)
        if scale == 1.0:
            return img

        hnew = min(round(himg * scale), hbound)
        wnew = min(round(wimg * scale), wbound)

        img = TF.resize(img, (hnew, wnew), self.interpolation)

        if self.pad is None:
            return img

        hpad = hbound - hnew
        wpad = wbound - wnew

        tpad = hpad // 2
        bpad = hpad - tpad

        lpad = wpad // 2
        rpad = wpad - lpad

        return TF.pad(img, (lpad, tpad, rpad, bpad), self.pad)

    def __repr__(self) -> str:
        return (
            f"{self.__class__.__name__}(" +
            f"bounds={self.bounds}, " +
            f"interpolation={self.interpolation.value}, " +
            f"grow={self.grow}, " +
            f"pad={self.pad})"
        )

class CompositeAlpha(torch.nn.Module):
    def __init__(
        self,
        background: tuple[float, float, float] | float,
    ):
        super().__init__()

        self.background = (background, background, background) if isinstance(background, float) else background
        self.background = torch.tensor(self.background).unsqueeze(1).unsqueeze(2)

    def forward(self, img: torch.Tensor) -> torch.Tensor:
        if img.shape[-3] == 3:
            return img

        alpha = img[..., 3, None, :, :]

        img[..., :3, :, :] *= alpha

        background = self.background.expand(-1, img.shape[-2], img.shape[-1])
        if background.ndim == 1:
            background = background[:, None, None]
        elif background.ndim == 2:
            background = background[None, :, :]

        img[..., :3, :, :] += (1.0 - alpha) * background
        return img[..., :3, :, :]

    def __repr__(self) -> str:
        return (
            f"{self.__class__.__name__}(" +
            f"background={self.background})"
        )

transform = transforms.Compose([
    Fit((384, 384)),
    transforms.ToTensor(),
    CompositeAlpha(0.5),
    transforms.Normalize(mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5], inplace=True),
    transforms.CenterCrop((384, 384)),
])

model = timm.create_model(
    "vit_so400m_patch14_siglip_384.webli",
    pretrained=False,
    num_classes=9083,
) # type: VisionTransformer

safetensors.torch.load_model(model, "JTP_PILOT/JTP_PILOT-e4-vit_so400m_patch14_siglip_384.safetensors")
model.eval()

if torch.cuda.is_available():
    model.cuda()
    if torch.cuda.get_device_capability()[0] >= 7: # tensor cores
        model.to(dtype=torch.float16, memory_format=torch.channels_last)

with open("JTP_PILOT/tags.json", "r") as file:
    tags = json.load(file) # type: dict
allowed_tags = list(tags.keys())

for idx, tag in enumerate(allowed_tags):
    allowed_tags[idx] = tag.replace("_", " ")

def create_tags(image, threshold):
    img = image.convert('RGB')
    tensor = transform(img).unsqueeze(0) # type: torch.Tensor

    if torch.cuda.is_available():
        tensor = tensor.cuda()
        if torch.cuda.get_device_capability()[0] >= 7:
            tensor = tensor.to(dtype=torch.float16, memory_format=torch.channels_last)

    with torch.no_grad():
        logits = model(tensor)
        probabilities = torch.nn.functional.sigmoid(logits[0])
        indices = torch.where(probabilities > threshold)[0]
        values = probabilities[indices]

    temp = []
    tag_score = dict()
    for i in range(indices.size(0)):
        temp.append([allowed_tags[indices[i]], values[i].item()])
        tag_score[allowed_tags[indices[i]]] = values[i].item()
    temp = [t[0] for t in temp]
    text_no_impl = ", ".join(temp)
    return text_no_impl, tag_score

with gr.Blocks() as demo:
    gr.Markdown("""
    ## Joint Tagger Project: PILOT
    This tagger is designed for use on furry images (though may very well work on out-of-distribution images, potentially with funny results).  A threshold of 0.2 is recommended.  Lower thresholds often turn up more valid tags, but can also result in some amount of hallucinated tags.

    This tagger is the result of joint efforts between members of the RedRocket team.

    Special thanks to Minotoro at frosting.ai for providing the compute power for this project.
    """)
    gr.Interface(
        create_tags,
        inputs=[gr.Image(label="Source", sources=['upload', 'webcam'], type='pil'), gr.Slider(minimum=0.00, maximum=1.00, step=0.01, value=0.20, label="Threshold")],
        outputs=[
            gr.Textbox(label="Tag String"),
            gr.Label(label="Tag Predictions", num_top_classes=200),
        ],
        allow_flagging="never",
    )

if __name__ == "__main__":
    demo.launch()