PokeDex-value error

#1

code is executing in jupyter but it training time is very long so i tried to execute on collab but when i use np.load it showing this error

cannot reshape array of size 2883552 into shape (1252,227,227,3)

#2

1252227227*3 = 193542924 which is not equal to 2883552, and hence the error is shown.

Please share your colab notebook link.

Moreover in the training data, images are correct till 328 only and after that there is random noise only and nothing else. So you are requested to take initial 328 images only as training data.

Hope that helped you :blush:

#3

i changed my input size to 328 it executed faster but

#4

changed my input size to 328 it executed faster but when i use predict there are lot value missing and np.load is not working on collab .i am attaching my collab link https://colab.research.google.com/drive/12k6By8iKukoQWH5LWR27HaJaFQ6aSJdr

#5

Your model predicts probabilty of each class so inverse transform is unable to apply inverse_transformation function.

Use np.argmax(y_pred, axis = 1) function to get class having maximum probability for model predictions.

Hope this helped :blush:

#6

i applied np.argmax but after 62 value are missing can you please check this i am attaching link for csv file and collabv link—csv :https://drive.google.com/file/d/1l_yCrTTv_JCZrDzJbT12nLDalW0Yc6Lc/view?usp=sharing collab: https://colab.research.google.com/drive/1NGjb3znCez5g7OSm_rc97rmkzn9ecoIJ

#7

Actually there is problem with test dataset as well, the values are not missing your model predicted class 0 because after 62 there are no actual images, there is problem in test data like in train dataset.

Happy Learning