debajyotikarmaker / cvpr_4234

49 stars 65 forks source link

Valid folder not found in the given dataset for PA_02 #14

Open MahmodulRatul opened 3 years ago

MahmodulRatul commented 3 years ago

In the dataset of following link, I did not found the valid folder. Only train and test is there. https://www.kaggle.com/grassknoted/asl-alphabet Anyone for guide me ?

SomyaDhar commented 3 years ago

Validation data did not given in kaggle .

But you can split validation data for train data set .

import this bellow code- from sklearn.model_selection import train_test_split and split by.........load all data in---- train_data levels = y_train for you............... x_train, x_valid, y_train, y_valid = train_test_split(train_data, labels, test_size = 0.1)

You can also give validation data size on .... test_size =0.2

.......................................

however, you can also give validation data in the fit method...... like ----validation_split=0.1

h = model.fit(x=x_train, y=y_train, epochs=20,validation_split=0.1, batch_size=128)

MahmodulRatul commented 3 years ago

Validation data did not given in kaggle .

But you can split validation data for train data set .

import this bellow code- from sklearn.model_selection import train_test_split and split by.........load all data in---- train_data levels = y_train for you............... x_train, x_valid, y_train, y_valid = train_test_split(train_data, labels, test_size = 0.1)

You can also give validation data size on .... test_size =0.2

.......................................

however, you can also give validation data in the fit method...... like ----validation_split=0.1

h = model.fit(x=x_train, y=y_train, epochs=20,validation_split=0.1, batch_size=128)

Uncountable thanks to you brother