Помогите найти ошибку )
Есть функция
def encode_single_sample(img_path, label):
# 1. Read image
img = tf.io.read_file(img_path)
# 2. Decode and convert to grayscale
img = tf.io.decode_png(img, channels=1)
# 3. Convert to float32 in [0, 1] range
img = tf.image.convert_image_dtype(img, tf.float32)
# 4. Resize to the desired size
img = tf.image.resize(img, [img_height, img_width])
# 5. Transpose the image because we want the time
# dimension to correspond to the width of the image.
img = tf.transpose(img, perm=[1, 0, 2])
# 6. Map the characters in label to numbers
label = char_to_num(tf.strings.unicode_split(label, input_encoding="UTF-8"))
# 7. Return a dict as our model is expecting two inputs
return {"image": img, "label": label}
Вызываю:
train_dataset = tf.data.Dataset.from_tensor_slices((x_train, y_train))
train_dataset = (
train_dataset.map(
encode_single_sample, num_parallel_calls=tf.data.AUTOTUNE
)
.batch(batch_size)
.prefetch(buffer_size=tf.data.AUTOTUNE)
)
Локально на маке все работает, python 3.9
На сервере с linux, python 3.10 вылетает ошибка:
TypeError: in user code:
File "/srv/solver/kdm/use_model.py", line 75, in encode_single_sample *
img = tf.io.read_file(img_path)
TypeError: Input 'filename' of 'ReadFile' Op has type float64 that does not match expected type of string.
UPD: Помимо этого там простое апи на flask и собственно когда я запускаю flask, он сыпется с этой ошибкой.
На всякий случай, flask запускаю так:
source venv/bin/activate &&
set -a &&
source .env.prod &&
set +a &&
flask --app api run --host=0.0.0.0 -p 5000
UPD2:
Проблема в типе данных:
def encode_single_sample(img_path, label):
# 1. Read image
print("img_path", img_path, type(img_path))
# print("label", img_path, type(label))
# if type(img_path) is not str:
# img_path = str(img_path)
# if type(label) is not str:
# label = str(label)
img = tf.io.read_file(img_path)
# 2. Decode and convert to grayscale
img = tf.io.decode_png(img, channels=1)
# 3. Convert to float32 in [0, 1] range
img = tf.image.convert_image_dtype(img, tf.float32)
# 4. Resize to the desired size
img = tf.image.resize(img, [img_height, img_width])
# 5. Transpose the image because we want the time
# dimension to correspond to the width of the image.
img = tf.transpose(img, perm=[1, 0, 2])
# 6. Map the characters in label to numbers
label = char_to_num(tf.strings.unicode_split(label, input_encoding="UTF-8"))
# 7. Return a dict as our model is expecting two inputs
return {"image": img, "label": label}
на линуксе:
print("img_path", img_path, type(img_path)) :
img_path Tensor("args_0:0", shape=(), dtype=float6
4) <class 'tensorflow.python.framework.ops.SymbolicTensor'>
на маке:
img_path Tensor("args_0:0", shape=(), dtype=string) <class 'tensorflow.python.framework.ops.SymbolicTensor'>
