1 | initial version |

As @berak point out that suspect. I was looking at my code. The suspect is **image** in parameter. You can't used image in `numpy`

array. You usually used index array. I have snippet code:

```
m1 = [270, 110, 150, 120, 370, 570, 110, 290, 350,380]
m2 = [55, 55, 51, 92, 78, 27, 14, 8, 26,54]
m3 = [38, 35, 39, 98, 57, 62, 10, 62, 43,89]
n1 = [0, 0, 1]
n2 = [0, 1, 0]
n3 = [1, 0, 0]
SAMPLES = 5000
for x in range(0, SAMPLES):
print(f'Samples {}{}'.format(x, SAMPLES)
ann.train(np.array([m1, m2, im3], np.float32),
cv2.ml.ROW_SAMPLE,
np.array([n1, n2, n3], np.float32))
```

2 | No.2 Revision |

As @berak point out that suspect. I was looking at my code. The suspect is **image** in parameter. You can't used image in `numpy`

array. You usually used index array. I have snippet code:

```
m1 = [270, 110, 150, 120, 370, 570, 110, 290, 350,380]
m2 = [55, 55, 51, 92, 78, 27, 14, 8, 26,54]
m3 = [38, 35, 39, 98, 57, 62, 10, 62, 43,89]
n1 = [0, 0, 1]
n2 = [0, 1, 0]
n3 = [1, 0, 0]
SAMPLES = 5000
for x in range(0, SAMPLES):
print(f'Samples {}{}'.format(x, SAMPLES)
ann.train(np.array([m1, m2,
```~~im3], ~~m3], np.float32),
cv2.ml.ROW_SAMPLE,
np.array([n1, n2, n3], np.float32))

3 | No.3 Revision |

As @berak point out that suspect. I was looking at my code. The suspect is **image** in parameter. You can't used image in `numpy`

array. You usually used index array. I have snippet code:

```
m1 = [270, 110, 150, 120, 370, 570, 110, 290, 350,380]
m2 = [55, 55, 51, 92, 78, 27, 14, 8, 26,54]
m3 = [38, 35, 39, 98, 57, 62, 10, 62, 43,89]
n1 = [0, 0, 1]
n2 = [0, 1, 0]
n3 = [1, 0, 0]
SAMPLES = 5000
for x in range(0, SAMPLES):
print(f'Samples {}{}'.format(x, SAMPLES)
ann.train(np.array([m1, m2, m3], np.float32),
cv2.ml.ROW_SAMPLE,
np.array([n1, n2, n3], np.float32))
```

Edit: As @berak point out w//out looping. I was doing `tensorflow`

and `keras`

. You can do like this:

```
ann.train(np.array([m1, m2, m3], np.float32),
cv2.ml.ROW_SAMPLE,
np.array([n1, n2, n3], np.float32))
```

Copyright OpenCV foundation, 2012-2018. Content on this site is licensed under a Creative Commons Attribution Share Alike 3.0 license.