Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -1,74 +1,32 @@
|
|
1 |
-
from fastapi import FastAPI, File, UploadFile, HTTPException
|
2 |
-
from PIL import Image
|
3 |
-
import numpy as np
|
4 |
import pickle
|
5 |
-
from
|
|
|
|
|
|
|
|
|
6 |
import math
|
7 |
|
8 |
-
|
9 |
-
|
10 |
-
# Cargar el modelo SOM previamente entrenado
|
11 |
-
with open("som.pkl", "rb") as tf:
|
12 |
-
som = pickle.load(tf)
|
13 |
-
|
14 |
-
M = np.array([
|
15 |
-
[ 0., -1., -1., -1., -1., 2., -1., -1., -1., 3.],
|
16 |
-
[-1., -1., -1., -1., -1., -1., -1., -1., -1., -1.],
|
17 |
-
[-1., -1., -1., 1., -1., -1., -1., -1., -1., -1.],
|
18 |
-
[ 1., -1., -1., -1., -1., -1., -1., -1., -1., 0.],
|
19 |
-
[-1., -1., -1., -1., 1., -1., -1., -1., -1., -1.],
|
20 |
-
[-1., -1., -1., -1., -1., -1., -1., -1., -1., -1.],
|
21 |
-
[ 3., -1., -1., -1., -1., -1., -1., -1., -1., 3.],
|
22 |
-
[-1., -1., -1., 0., -1., -1., 3., -1., -1., -1.],
|
23 |
-
[-1., -1., -1., -1., -1., -1., -1., -1., -1., -1.],
|
24 |
-
[ 2., -1., -1., -1., 1., -1., -1., -1., -1., 2.]
|
25 |
-
])
|
26 |
-
|
27 |
-
def predict_fingerprint(image):
|
28 |
-
try:
|
29 |
-
processed_image = preprocess_image(image)
|
30 |
-
winner = som.winner(processed_image)
|
31 |
-
fingerprint_type = get_fingerprint_type(winner)
|
32 |
-
return fingerprint_type
|
33 |
-
except Exception as e:
|
34 |
-
raise HTTPException(status_code=500, detail=str(e))
|
35 |
|
36 |
-
|
37 |
-
# Guardar la imagen en formato TIFF
|
38 |
-
processed_image = representativo(image)
|
39 |
-
processed_image_resized = processed_image.reshape(1, -1)
|
40 |
-
print(f"\n\n\nXD:{processed_image.size}\t{processed_image_resized.size}")
|
41 |
-
return processed_image_resized
|
42 |
-
|
43 |
-
def get_fingerprint_type(winner):
|
44 |
-
labels = {0: "LL", 1: "RL", 2: "WH", 3: "AR"}
|
45 |
-
fingerprint_type = labels[int(M[winner[0], winner[1]])]
|
46 |
-
return fingerprint_type
|
47 |
|
48 |
-
|
49 |
-
|
50 |
-
|
51 |
-
|
52 |
-
|
53 |
-
image.save("temp_image.tif")
|
54 |
-
image = Image.open("temp_image.tif")
|
55 |
-
image.resize((256,256))
|
56 |
-
print(f"\n\n\n\nSIZE:{image.size}\t{image}")
|
57 |
-
fingerprint_type = predict_fingerprint(image)
|
58 |
-
return {"prediction": fingerprint_type}
|
59 |
-
except Exception as e:
|
60 |
-
raise HTTPException(status_code=500, detail=str(e))
|
61 |
|
62 |
def sobel(I):
|
63 |
m, n = I.shape
|
64 |
Gx = np.zeros([m-2, n-2], np.float32)
|
65 |
Gy = np.zeros([m-2, n-2], np.float32)
|
66 |
-
gx =
|
67 |
-
gy =
|
68 |
for j in range(1, m-2):
|
69 |
for i in range(1, n-2):
|
70 |
-
Gx[j-1, i-1] =
|
71 |
-
Gy[j-1, i-1] =
|
72 |
return Gx, Gy
|
73 |
|
74 |
def medfilt2(G, d=3):
|
@@ -78,8 +36,8 @@ def medfilt2(G, d=3):
|
|
78 |
temp[1:m+1, 1:n+1] = G
|
79 |
for i in range(1, m):
|
80 |
for j in range(1, n):
|
81 |
-
A = np.
|
82 |
-
salida[i-1, j-1] = A[d+1]
|
83 |
return salida
|
84 |
|
85 |
def orientacion(patron, w):
|
@@ -90,14 +48,43 @@ def orientacion(patron, w):
|
|
90 |
mOrientaciones = np.zeros([m//w, n//w], np.float32)
|
91 |
for i in range(m//w):
|
92 |
for j in range(n//w):
|
93 |
-
YY =
|
94 |
-
XX =
|
95 |
-
mOrientaciones[i, j] = (0.5 * math.atan2(YY, XX) + math.pi/2.0) * (180.0 / math.pi)
|
96 |
return mOrientaciones
|
97 |
|
98 |
-
def representativo(
|
99 |
-
|
100 |
-
|
101 |
patron = imarray[1:m-1, 1:n-1]
|
102 |
EE = orientacion(patron, 14)
|
103 |
-
return np.asarray(EE).reshape(-1)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
import pickle
|
2 |
+
from minisom import MiniSom
|
3 |
+
import numpy as np
|
4 |
+
from fastapi import FastAPI, HTTPException
|
5 |
+
from pydantic import BaseModel
|
6 |
+
from typing import List
|
7 |
import math
|
8 |
|
9 |
+
class InputData(BaseModel):
|
10 |
+
array: List[List[int]]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
11 |
|
12 |
+
app = FastAPI()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
13 |
|
14 |
+
# Cargar el modelo SOM
|
15 |
+
def load_model():
|
16 |
+
with open('som.pkl', 'rb') as fid:
|
17 |
+
somecoli = pickle.load(fid)
|
18 |
+
return somecoli
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
19 |
|
20 |
def sobel(I):
|
21 |
m, n = I.shape
|
22 |
Gx = np.zeros([m-2, n-2], np.float32)
|
23 |
Gy = np.zeros([m-2, n-2], np.float32)
|
24 |
+
gx = [[-1, 0, 1], [-2, 0, 2], [-1, 0, 1]]
|
25 |
+
gy = [[1, 2, 1], [0, 0, 0], [-1, -2, -1]]
|
26 |
for j in range(1, m-2):
|
27 |
for i in range(1, n-2):
|
28 |
+
Gx[j-1, i-1] = sum(sum(I[j-1:j+2, i-1:i+2] * gx))
|
29 |
+
Gy[j-1, i-1] = sum(sum(I[j-1:j+2, i-1:i+2] * gy))
|
30 |
return Gx, Gy
|
31 |
|
32 |
def medfilt2(G, d=3):
|
|
|
36 |
temp[1:m+1, 1:n+1] = G
|
37 |
for i in range(1, m):
|
38 |
for j in range(1, n):
|
39 |
+
A = np.asarray(temp[i-1:i+2, j-1:j+2]).reshape(-1)
|
40 |
+
salida[i-1, j-1] = np.sort(A)[d+1]
|
41 |
return salida
|
42 |
|
43 |
def orientacion(patron, w):
|
|
|
48 |
mOrientaciones = np.zeros([m//w, n//w], np.float32)
|
49 |
for i in range(m//w):
|
50 |
for j in range(n//w):
|
51 |
+
YY = sum(sum(2*Gx[i*w:(i+1)*w, j:j+1] * Gy[i*w:(i+1)*w, j:j+1]))
|
52 |
+
XX = sum(sum(Gx[i*w:(i+1)*w, j:j+1]**2 - Gy[i*w:(i+1)*w, j:j+1]**2))
|
53 |
+
mOrientaciones[i, j] = (0.5 * math.atan2(YY, XX) + math.pi / 2.0) * (180.0 / math.pi)
|
54 |
return mOrientaciones
|
55 |
|
56 |
+
def representativo(imarray):
|
57 |
+
imarray = np.squeeze(imarray)
|
58 |
+
m, n = imarray.shape
|
59 |
patron = imarray[1:m-1, 1:n-1]
|
60 |
EE = orientacion(patron, 14)
|
61 |
+
return np.asarray(EE).reshape(-1)
|
62 |
+
|
63 |
+
som = load_model()
|
64 |
+
|
65 |
+
MM = np.array([
|
66 |
+
[ 0., -1., -1., -1., -1., 2., -1., -1., -1., 3.],
|
67 |
+
[-1., -1., -1., -1., -1., -1., -1., -1., -1., -1.],
|
68 |
+
[-1., -1., -1., 1., -1., -1., -1., -1., -1., -1.],
|
69 |
+
[ 1., -1., -1., -1., -1., -1., -1., -1., -1., 0.],
|
70 |
+
[-1., -1., -1., -1., 1., -1., -1., -1., -1., -1.],
|
71 |
+
[-1., -1., -1., -1., -1., -1., -1., -1., -1., -1.],
|
72 |
+
[ 3., -1., -1., -1., -1., -1., -1., -1., -1., 3.],
|
73 |
+
[-1., -1., -1., 0., -1., -1., 3., -1., -1., -1.],
|
74 |
+
[-1., -1., -1., -1., -1., -1., -1., -1., -1., -1.],
|
75 |
+
[ 2., -1., -1., -1., 1., -1., -1., -1., -1., 2.]
|
76 |
+
])
|
77 |
+
|
78 |
+
@app.post("/predict/")
|
79 |
+
async def predict(data: InputData):
|
80 |
+
try:
|
81 |
+
input_data = np.array(data.array).reshape(256, 256, 1)
|
82 |
+
representative_data = representativo(input_data)
|
83 |
+
representative_data = representative_data.reshape(1, -1)
|
84 |
+
|
85 |
+
w = som.winner(representative_data)
|
86 |
+
prediction = MM[w]
|
87 |
+
|
88 |
+
return {"prediction": prediction}
|
89 |
+
except Exception as e:
|
90 |
+
raise HTTPException(status_code=500, detail=str(e))
|