how to add more data with mnist data to train keras model

Question

I tried to recognize using mnist data as train data but failed so I want to add my train data with that so how can I train my model with mnist dataset and my own data. I tried training with my own data also but failed any suggestions are welcome ..... for example first I trained with mnist dataset followed this approach

https://medium.com/coinmonks/handwritten-digit-prediction-using-convolutional-neural-networks-in-tensorflow-with-keras-and-live-5ebddf46dc8

but the results are not satisfying so I tried to train that model by my data but looking at the results maybe I don't have enough data so now I want to train my model using mnist data + my data . so how can I do that

To give you a great answer, it might help us if you have a glance at [ask] if you haven't already. It might be also useful if you could provide a [mcve]. — Mat, Oct 22 '18 at 13:00
I tried explaining a little more please help me out here @Mat — kritarth, Oct 22 '18 at 13:23

score 0 · Answer 1 · answered Oct 23 '18 at 07:52

There are several ways how to concatenate two numpy arrays. The most obvious one is probably np.concatenate()

import numpy as np

all_train_x = np.concatenate((mnist_train_x, my_train_x))
all_train_y = np.concatenate((mnist_train_y, my_train_y))

Note the double parentheses.

score 0 · Answer 2 · answered Oct 23 '18 at 08:16

0

You can also try data augmentation technique. This will increase you size of the dataset and will also improve the performance of your model. You can some useful information here.

answered Oct 23 '18 at 08:16

Akhilesh Pandey

868
8
18

how to add more data with mnist data to train keras model

2 Answers2