python - 制作cnn+lstm模型时layer time_distributed的error input 0 is in compatible with the layer

Question

再会，

我正在尝试使用我自己的数据从头开始制作 cnn(Alexnet)+lstm

结果当我运行model.fit时出现了这个错误

ValueError: 层 time_distributed_15 的输入 0 与层不兼容：预期 ndim=5，发现 ndim=4。收到的完整形状：（无，无，无，无）

我的模型摘要如下

Model: "model_3"
_________________________________________________________________ Layer (type)                 Output Shape              Param #   
================================================================= input_4 (InputLayer)         [(None, 10, 224, 224, 3)] 0         
_________________________________________________________________ time_distributed_15 (TimeDis (None, 10, 56, 56, 96)    34944     
_________________________________________________________________ time_distributed_16 (TimeDis (None, 10, 56, 56, 96)    384       
_________________________________________________________________ time_distributed_17 (TimeDis (None, 10, 28, 28, 96)    0         
_________________________________________________________________ time_distributed_18 (TimeDis (None, 10, 14, 14, 256)   614656    
_________________________________________________________________ time_distributed_19 (TimeDis (None, 10, 14, 14, 256)   1024      
_________________________________________________________________ time_distributed_20 (TimeDis (None, 10, 7, 7, 256)     0         
_________________________________________________________________ time_distributed_21 (TimeDis (None, 10, 7, 7, 384)     885120    
_________________________________________________________________ time_distributed_22 (TimeDis (None, 10, 7, 7, 384)     1536      
_________________________________________________________________ time_distributed_23 (TimeDis (None, 10, 7, 7, 384)     1327488   
_________________________________________________________________ time_distributed_24 (TimeDis (None, 10, 7, 7, 384)     1536      
_________________________________________________________________ time_distributed_25 (TimeDis (None, 10, 7, 7, 256)     884992    
_________________________________________________________________ time_distributed_26 (TimeDis (None, 10, 7, 7, 256)     1024      
_________________________________________________________________ time_distributed_27 (TimeDis (None, 10, 3, 3, 256)     0         
_________________________________________________________________ time_distributed_28 (TimeDis (None, 10, 2304)          0         
_________________________________________________________________ lstm_3 (LSTM)                (None, 10, 32)            299136    
_________________________________________________________________ lstm_4 (LSTM)                (None, 32)                8320      
_________________________________________________________________ dense_7 (Dense)              (None, 1024)              33792     
_________________________________________________________________ batch_normalization_21 (Batc (None, 1024)              4096      
_________________________________________________________________ dropout_4 (Dropout)          (None, 1024)              0         
_________________________________________________________________ dense_8 (Dense)              (None, 512)               524800    
_________________________________________________________________ dropout_5 (Dropout)          (None, 512)               0         
_________________________________________________________________ dense_9 (Dense)              (None, 64)                32832     
_________________________________________________________________ dropout_6 (Dropout)          (None, 64)                0         
_________________________________________________________________ dense_10 (Dense)             (None, 6)                 390       
================================================================= Total params: 4,656,070 Trainable params: 4,651,270 Non-trainable params: 4,800

---

这是我如何制作模型的代码

https://colab.research.google.com/drive/12KChNk0t2NdFZA1_8H0nZ-5HYubPqAr9?usp=sharing

以及我如何放置我的数据集树是这样的

Sortir2
 -Class1
   -img1.jpg
   -img2.jpg
   -img3.jpg
 -Class2
   -img4.jpg
   -img5.jpg
   -img6.jpg

我在互联网上搜索了一些解决方案，但它仍然不起作用

我希望这个领域的任何人（深度学习）或了解如何解决这个问题的人可以帮助我

太感谢了

标签： pythondeep-learningconv-neural-networklstm