我怎样才能解决 not an sstable bad magic number 错误

Question

我和我的朋友正在为黑客马拉松制作图像识别的深度学习模型，我们不断遇到这个问题。

基本上，当我运行 run.py 进行分析和成像时，它会返回 sstable（坏幻数）错误。

我们不知道为什么会这样，也不知道该怎么办。

这是 run.py：

    import os, gc

from skimage import io

import glob

import pandas as pd

import glob

import tensorflow as tf

from tensorflow import keras

from keras.preprocessing import image

from tensorflow.keras.models import Sequential, save_model, load_model

import matplotlib.pyplot as plt

import numpy as np

from skimage import transform

from keras.optimizers import Adam

from keras.applications import mobilenet_v2

from PIL import Image

path = []

for file in os.listdir("./media_cdn"):

    path.append(file)

print(path)

filepath = './saved_model'

model = load_model(filepath, custom_objects= None, compile = False)

loss = 'CategoricalCrossentropy'

optimizer = Adam(lr=1e-5)

metrics = ['binary_accuracy']

model.compile(optimizer=optimizer, loss=loss, metrics=metrics)

def load(filename):

np_image = Image.open("./media_cdn/" + filename)

np_image = np.array(np_image).astype('float32')/255

np_image = transform.resize(np_image, (244, 244, 3))

np_image = np.expand_dims(np_image, axis=0)

return np_image

new_image = load(path[0])

print(new_image.shape)

new_model = keras.Sequential([model])

new_model.load_weights('./model_weights')

prediction = new_model.predict_classes(new_image)

classes = np.argmax(prediction, axis = -1)

print(classes)

print('This is the Diagnosis:')

if classes == 0:

    print('MELANOMA')

if classes == 1:

    print('Melanocytic Nevus')

if classes == 2:

    print('Basal Cell Carcinoma')

if classes == 3:

    print('Arctinic Keratosis')

if classes == 4:

    print('Benign Keratosis')

if classes == 5:

    print('Dermatofibroma')

if classes == 6:

    print('Vascular Lesion')

if classes == 7:

    print('Squamous Cell Carcinoma')

if classes == 8:

    print(['Unknown', 'BCC', 'AK', 'BKL', 'DF', 'VASC', 'SCC', 'UNK'])

classes = np.argmax(prediction, axis = 1)

print(classes)

调试时，错误显示在行中load_model。

我们不知道如何修复它，欢迎任何帮助。

Answer 1

好的，我知道为什么会发生这种情况。看来我必须在我的计算机上运行该模型，以便它可以生成正确的变量和模型文件。