forked from stsievert/next-face-demo
-
Notifications
You must be signed in to change notification settings - Fork 0
/
train.py
64 lines (52 loc) · 1.64 KB
/
train.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
import json
import pickle
import numpy as np
import pandas as pd
from joblib import Parallel, delayed
from sklearn.kernel_ridge import KernelRidge
from sklearn.model_selection import train_test_split
import face_api
DIR = "./train-files/"
y = pd.read_csv(DIR + "embedding.csv", index_col="Item")
if False:
D = {
face: face_api.distances("faces/" + face + ".png")
for face in y.index
if face not in {"40M_AN_O"}
}
with open(DIR + "D.pkl", "wb") as f:
pickle.dump(D, f)
else:
with open(DIR + "D.pkl", "rb") as f:
D = pickle.load(f)
y = dict(y.T)
y = {key: value for key, value in y.items() if key in D.keys()}
y = pd.DataFrame(y).T
D = pd.DataFrame(D).T
assert len(D) == len(y)
len_before = len(D)
cols = D.columns
df = pd.merge(y, D, left_index=True, right_index=True)
y = df[["x", "y"]]
D = df[cols]
assert len(D) == len_before
assert all(y.index == D.index), "Make sure the features and items are ordered!"
y = y.values
D = D.values
n, p = len(y), D.shape[1]
train, test = train_test_split(np.arange(n), test_size=0.1)
model = KernelRidge(alpha=1 / n)
model.fit(D[train], y[train])
np.savez(DIR + "features_and_embedding.npyz", y=y, D=D)
if __name__ == "__main__":
y_hat = model.predict(D[test])
distance = np.linalg.norm(y_hat - y[test], axis=1)
distance /= y.max() - y.min()
print("median distance", np.median(distance))
import matplotlib.pyplot as plt
plt.figure()
for y_h, yi in zip(y_hat, y):
plt.plot(*y_h, "ro")
plt.plot(*yi, "bo")
plt.plot([yi[0], y_h[0]], [yi[1], y_h[1]], "y--")
plt.show()