IndexError：在model.fit（）中列出超出范圍的索引

Question

我是使用 tensorflow 的新手。 我正在嘗試用形狀 (16*16) 的圖像訓練我的網絡。 我將 512*512 的 3 張灰度圖像分成 16*16 並全部附加。 所以我有 3072*16*16。 在訓練時我遇到錯誤。 我正在使用 jupyter notebook。誰能幫幫我？

這是代碼

import tensorflow as tf 
import numpy as np
from numpy import newaxis
import glob
import os
from PIL import Image,ImageOps
import random
from os.path import join
import matplotlib.pyplot as plt
from tensorflow import keras
TRAIN_PATH = 'dataset/2/*.jpg'
LOGS_Path = "dataset/logs/"
CHECKPOINTS_PATH = 'dataset/checkpoints/'
BETA = .75
EXP_NAME = f"beta_{BETA}"

files_list = glob.glob(join(TRAIN_PATH))
leng=len(files_list)
new_cover = []
for i in range(leng):
    img_cover_path = files_list[i]   
    for j in range (0,512,16):
        for k in range (0,512,16):
        img_cover = Image.open(img_cover_path)
        area=(k,j,k+16,j+16)
        img_cover1=img_cover.crop(area)
        img_cover1 = np.array(ImageOps.fit(img_cover1(16,16)),dtype=np.float32)
        img_cover1 /= 255.
        n1.append(img_cover1)


    new_cover.append(n1)


new_cover = np.array(new_cover) 
new_cover1=np.swapaxes(new_cover, 1,3) 

tf.reset_default_graph()
model=keras.Sequential()

#1st
model.add(keras.layers.Conv2D(64, (3, 3), strides=1,padding='SAME', input_shape = (16, 16, 3072))) #number of filters,shape of filter,input image size,activation function
model.add(keras.layers.BatchNormalization())
model.add(keras.layers.Activation(activation='relu'))
#2
model.add(keras.layers.Conv2D(64, (3, 3),strides=1,padding='SAME')) #number of filters,shape of filter,input image size,activation function
model.add(keras.layers.BatchNormalization())
model.add(keras.layers.Activation(activation='relu'))
#3
model.add(keras.layers.Conv2D(64, (3, 3),strides=1,padding='SAME')) #number of filters,shape of filter,input image size,activation function
model.add(keras.layers.BatchNormalization())
model.add(keras.layers.Activation(activation='relu'))
#4
model.add(keras.layers.Conv2D(64, (3, 3),strides=1,padding='SAME')) #number of filters,shape of filter,input image size,activation function
model.add(keras.layers.BatchNormalization())
model.add(keras.layers.Activation(activation='relu'))
#message

#compiling
model.compile(optimizer = tf.train.AdamOptimizer(0.001),loss='mse', metrics = ['accuracy'])
model.summary()
# Store training stats
model.fit(x=new_cover1,y=None, batch_size=32, epochs=1, verbose=1, callbacks=None, validation_split=0, validation_data=None, shuffle=True, class_weight=None, sample_weight=None, initial_epoch=0, steps_per_epoch=None, validation_steps=None)

它給出了錯誤：

_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
conv2d (Conv2D)              (None, 16, 16, 64)        1769536   
_________________________________________________________________
batch_normalization (BatchNo (None, 16, 16, 64)        256       
_________________________________________________________________
activation (Activation)      (None, 16, 16, 64)        0         
_________________________________________________________________
conv2d_1 (Conv2D)            (None, 16, 16, 64)        36928     
_________________________________________________________________
batch_normalization_1 (Batch (None, 16, 16, 64)        256       
_________________________________________________________________
activation_1 (Activation)    (None, 16, 16, 64)        0         
_________________________________________________________________
conv2d_2 (Conv2D)            (None, 16, 16, 64)        36928     
_________________________________________________________________
batch_normalization_2 (Batch (None, 16, 16, 64)        256       
_________________________________________________________________
activation_2 (Activation)    (None, 16, 16, 64)        0         
_________________________________________________________________
conv2d_3 (Conv2D)            (None, 16, 16, 64)        36928     
_________________________________________________________________
batch_normalization_3 (Batch (None, 16, 16, 64)        256       
_________________________________________________________________
activation_3 (Activation)    (None, 16, 16, 64)        0         
=================================================================
Total params: 1,881,344
Trainable params: 1,880,832
Non-trainable params: 512
_________________________________________________________________

---------------------------------------------------------------------------
IndexError                                Traceback (most recent call last)
<ipython-input-20-49da746cee1b> in <module>()
     24 model.summary()
     25 # Store training stats
---> 26 model.fit(x=new_cover1,y=None, batch_size=32, epochs=1, verbose=1, callbacks=None, validation_split=0, validation_data=None, shuffle=True, class_weight=None, sample_weight=None, initial_epoch=0, steps_per_epoch=None, validation_steps=None)
     27 
     28 #return model

~\AppData\Local\Continuum\anaconda3\lib\site-packages\tensorflow\python\keras\engine\training.py in fit(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, max_queue_size, workers, use_multiprocessing, **kwargs)
   1654           initial_epoch=initial_epoch,
   1655           steps_per_epoch=steps_per_epoch,
-> 1656           validation_steps=validation_steps)
   1657 
   1658   def evaluate(self,

~\AppData\Local\Continuum\anaconda3\lib\site-packages\tensorflow\python\keras\engine\training_arrays.py in fit_loop(model, inputs, targets, sample_weights, batch_size, epochs, verbose, callbacks, val_inputs, val_targets, val_sample_weights, shuffle, initial_epoch, steps_per_epoch, validation_steps)
    135   indices_for_conversion_to_dense = []
    136   for i in range(len(feed)):
--> 137     if issparse is not None and issparse(ins[i]) and not K.is_sparse(feed[i]):
    138       indices_for_conversion_to_dense.append(i)
    139 

IndexError: list index out of range

Answer 1

經過我的研究，我發現這是一個已知問題，官方解決方案尚未發布。 盡管如此，仍然有一些建議聲稱有效。

建議更新到更新夜間構建版本（ pip install tf-nightly或pip install tf-nightly-gpu ）

https://github.com/tensorflow/tensorflow/issues/21894#issuecomment-418552609

Answer 2

我認為這個錯誤是因為傳遞給模型的 x 和 y 的形狀。 你通過 None 作為標簽！