This is an example of the seq2seq addition task on the LSTM model, that implemented on images
We've got label 7 or 8(in other examples) of the plus sign because I didn't train plus sign on the model, but I fixed it with a simple math trick
Run:
python main.py -i [IMAGE PATH] (in this case is ./images/[IMAGENAME.JPG])
Result would be:
[INFO] 4 - 91.32%
[INFO] 8 - 96.93%
[INFO] 3 - 99.95%
[INFO] 7 - 92.68%
[INFO] 2 - 89.00%
[INFO] 2 - 82.98%
Your operation is: 483+22
Result: 483+22 = 505