new dataset

[HanYuanyuaner]

@taki0112 I want to train new dataset with this model, must I resize the image to 32*32?

[jerryhero]

I have the same question as you,have you found way to train own pictures dataset? @HanYuanyuaner

[Ggmatch]

me too

[Bella722]

you could change prepare_data function to load your own dataset.

def prepare_data():
print("======Loading data======")
traindata_dir = 'e:/train/' #change your traindata_dir
testdata_dir = 'e:/test/' #change your testdata_dir
image_dim = image_size * image_size * img_channels
label_names = ['cardboard','glass','metal','trash','paper','plastic'] #change your classes
label_count = len(label_names)
train_files = [traindata_dir+s for s in label_names]
test_files = [testdata_dir+s for s in label_names]
train_data, train_labels = load_data(train_files, traindata_dir, label_count)
test_data, test_labels = load_data(test_files, testdata_dir, label_count)

print("Train data:", np.shape(train_data), np.shape(train_labels)) 
print("Test data :", np.shape(test_data), np.shape(test_labels))
print("======Load finished======") 
print("======Shuffling data======") 
indices = np.random.permutation(len(train_data)) 
train_data = train_data[indices] 
train_labels = train_labels[indices]
indices = np.random.permutation(len(test_data)) 
test_data = test_data[indices] 
test_labels = test_labels[indices]
print("======Prepare Finished======") 
return train_data, train_labels, test_data, test_labels

or you can train and test ratio to generate trainfile and testflie list then load.