-
Notifications
You must be signed in to change notification settings - Fork 0
/
openai_utils.py
75 lines (47 loc) · 1.7 KB
/
openai_utils.py
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
import clip
import torch
def get_model_slug_for_clip():
return "clip"
def count_num_features_for_clip(model=None):
num_features = 512
return num_features
def get_model_resolution_for_clip(model=None):
resolution = 224 if model is None else model.input_resolution.item()
return resolution
def get_device():
device = "cuda" if torch.cuda.is_available() else "cpu"
return device
def get_clip_model_name():
model_name = "ViT-B/32"
return model_name
def load_clip_tools():
model, preprocess = clip.load(get_clip_model_name(), device=get_device())
return model, preprocess
def get_preprocessing_for_clip():
_, preprocess = load_clip_tools()
return preprocess
def preprocess_image_array_for_model_for_clip(image_array, preprocess=None):
if preprocess is None:
preprocess = get_preprocessing_for_clip()
image_array = preprocess(image_array).unsqueeze(0)
return image_array
def get_model_for_clip(input_shape=None, include_top=None, pooling=None):
model, _ = load_clip_tools()
return model
def label_image_for_clip(image, model=None, preprocess=None, normalize_features=True):
if model is None:
model = get_model_for_clip()
preprocessed_image = preprocess_image_array_for_model_for_clip(
image,
preprocess=preprocess,
)
image_array = preprocessed_image.to(get_device())
with torch.no_grad():
image_features = model.encode_image(image_array)
if normalize_features:
image_features /= image_features.norm(dim=-1, keepdim=True)
yhat = image_features.cpu().numpy()
return yhat
if __name__ == "__main__":
slug_name = get_model_slug_for_clip()
print(f"Slug: {slug_name}")