Spaces:
Sleeping
Sleeping
Commit
·
c4ccf03
1
Parent(s):
53dc186
first commit
Browse files
app.py
CHANGED
@@ -6,13 +6,12 @@ import matplotlib.pyplot as plt
|
|
6 |
import requests
|
7 |
import os
|
8 |
|
9 |
-
#
|
|
|
|
|
10 |
MODEL_URL = "https://huggingface.co/nasa-ibm-ai4science/Surya-1.0/resolve/main/surya.366m.v1.pt"
|
11 |
-
|
12 |
-
# Nome local do arquivo
|
13 |
MODEL_FILE = "surya.366m.v1.pt"
|
14 |
|
15 |
-
# Função para baixar o modelo se não existir
|
16 |
def download_model():
|
17 |
if not os.path.exists(MODEL_FILE):
|
18 |
print("Baixando pesos do Surya-1.0...")
|
@@ -21,33 +20,52 @@ def download_model():
|
|
21 |
f.write(r.content)
|
22 |
print("Download concluído!")
|
23 |
|
24 |
-
# Baixar modelo
|
25 |
download_model()
|
26 |
|
27 |
-
#
|
28 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
29 |
model.eval()
|
30 |
|
31 |
-
#
|
|
|
|
|
32 |
def infer_solar_image_heatmap(img):
|
33 |
-
# Pré-processamento: grayscale, resize 224x224
|
34 |
img = img.convert("L").resize((224, 224))
|
35 |
img_tensor = torch.tensor(np.array(img), dtype=torch.float32).unsqueeze(0).unsqueeze(0) / 255.0
|
36 |
|
37 |
with torch.no_grad():
|
38 |
outputs = model(img_tensor)
|
39 |
|
40 |
-
# Criar heatmap
|
41 |
emb = outputs.squeeze().numpy()
|
42 |
heatmap = emb - emb.min()
|
43 |
-
heatmap /= heatmap.max() + 1e-8
|
44 |
|
45 |
plt.imshow(heatmap, cmap='hot')
|
46 |
plt.axis('off')
|
47 |
plt.tight_layout()
|
48 |
return plt.gcf()
|
49 |
|
50 |
-
#
|
|
|
|
|
51 |
interface = gr.Interface(
|
52 |
fn=infer_solar_image_heatmap,
|
53 |
inputs=gr.Image(type="pil"),
|
|
|
6 |
import requests
|
7 |
import os
|
8 |
|
9 |
+
# ================================
|
10 |
+
# 1. Baixar pesos do Surya-1.0
|
11 |
+
# ================================
|
12 |
MODEL_URL = "https://huggingface.co/nasa-ibm-ai4science/Surya-1.0/resolve/main/surya.366m.v1.pt"
|
|
|
|
|
13 |
MODEL_FILE = "surya.366m.v1.pt"
|
14 |
|
|
|
15 |
def download_model():
|
16 |
if not os.path.exists(MODEL_FILE):
|
17 |
print("Baixando pesos do Surya-1.0...")
|
|
|
20 |
f.write(r.content)
|
21 |
print("Download concluído!")
|
22 |
|
|
|
23 |
download_model()
|
24 |
|
25 |
+
# ================================
|
26 |
+
# 2. Definir a arquitetura do Surya
|
27 |
+
# ================================
|
28 |
+
# Aqui você deve colar ou importar a classe SuryaModel do repo oficial
|
29 |
+
# Exemplo genérico:
|
30 |
+
import torch.nn as nn
|
31 |
+
|
32 |
+
class SuryaModel(nn.Module):
|
33 |
+
def __init__(self):
|
34 |
+
super().__init__()
|
35 |
+
self.conv = nn.Conv2d(1, 1, kernel_size=3, padding=1)
|
36 |
+
def forward(self, x):
|
37 |
+
return self.conv(x)
|
38 |
+
|
39 |
+
# ================================
|
40 |
+
# 3. Criar instância e carregar pesos
|
41 |
+
# ================================
|
42 |
+
model = SuryaModel()
|
43 |
+
state_dict = torch.load(MODEL_FILE, map_location=torch.device('cpu'))
|
44 |
+
model.load_state_dict(state_dict)
|
45 |
model.eval()
|
46 |
|
47 |
+
# ================================
|
48 |
+
# 4. Função de inferência para heatmap
|
49 |
+
# ================================
|
50 |
def infer_solar_image_heatmap(img):
|
|
|
51 |
img = img.convert("L").resize((224, 224))
|
52 |
img_tensor = torch.tensor(np.array(img), dtype=torch.float32).unsqueeze(0).unsqueeze(0) / 255.0
|
53 |
|
54 |
with torch.no_grad():
|
55 |
outputs = model(img_tensor)
|
56 |
|
|
|
57 |
emb = outputs.squeeze().numpy()
|
58 |
heatmap = emb - emb.min()
|
59 |
+
heatmap /= heatmap.max() + 1e-8
|
60 |
|
61 |
plt.imshow(heatmap, cmap='hot')
|
62 |
plt.axis('off')
|
63 |
plt.tight_layout()
|
64 |
return plt.gcf()
|
65 |
|
66 |
+
# ================================
|
67 |
+
# 5. Interface Gradio
|
68 |
+
# ================================
|
69 |
interface = gr.Interface(
|
70 |
fn=infer_solar_image_heatmap,
|
71 |
inputs=gr.Image(type="pil"),
|