def read_split(split):
test_fname = 'test'+str(split)+'.txt'
part_dat = False
df_test = create_df(os.path.join(datapath, test_fname),
img_path,
partial_dataset=part_dat,
seed=123)
return df_test
def read_split_to_dfs(split):
train_fname = 'train' + str(split) + '.txt'
test_fname = 'test' + str(split) + '.txt'
val_fname = 'val' + str(split) + '.txt'
part_dat = False
df_train = create_df(os.path.join(datapath, train_fname),
img_path,
partial_dataset=part_dat,
seed=123)
df_test = create_df(os.path.join(datapath, test_fname),
img_path,
partial_dataset=part_dat,
seed=123)
df_val = create_df(os.path.join(datapath, val_fname),
img_path,
partial_dataset=part_dat,
seed=123)
return df_train, df_test, df_val
#Load the ready-made splits
if platform.system() == 'Linux':
datapath = '/home/mikko/Documents/kandi/data/IDA/Separate lists with numbering/Machine learning splits'
img_path = '/home/mikko/Documents/kandi/data/IDA/Images/'
else:
datapath = 'C:\\koodia\\kandi\\FIN Benthic2\\IDA\\Separate lists with numbering\\Machine learning splits'
img_path = 'C:\\koodia\\kandi\\FIN Benthic2\\IDA\\Images\\'
split = 1
test_fname = 'test'+str(split)+'.txt'
part_dat = True
df_test = create_df(os.path.join(datapath, test_fname),
img_path,
partial_dataset=part_dat,
seed=123)
#%% Loading in memory
import tensorflow as tf
def create_tf_img(fname, imsize):
img = tf.io.read_file(fname)
img = tf.image.decode_jpeg(img, channels=3)
img = tf.image.convert_image_dtype(img, tf.float32)
return tf.image.resize(img, imsize[0:-1])
def load_images(df, img_path, imsize=(224,224,3)):
imglist = df.loc[:,"path"].tolist()
