Trying to understand the output of model.predict

Question

I am trying to predict the cat/dog image using the model.predict method. Since it is 2 class classifier I am getting an array of 2 values. As per my understanding, these values represent the probabilities to belong in each class (correct me if I am wrong). If it is so the probability must be summed to 1. But I am getting the same probability for both classes

Model History

Model: "sequential"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
flatten (Flatten)            multiple                  0         
_________________________________________________________________
dense (Dense)                multiple                  30848     
_________________________________________________________________
dropout (Dropout)            multiple                  0         
_________________________________________________________________
batch_normalization (BatchNo multiple                  512       
_________________________________________________________________
dense_1 (Dense)              multiple                  12900     
_________________________________________________________________
dropout_1 (Dropout)          multiple                  0         
_________________________________________________________________
batch_normalization_1 (Batch multiple                  400       
_________________________________________________________________
dense_2 (Dense)              multiple                  10100     
_________________________________________________________________
dropout_2 (Dropout)          multiple                  0         
_________________________________________________________________
batch_normalization_2 (Batch multiple                  400       
_________________________________________________________________
dense_3 (Dense)              multiple                  10100     
_________________________________________________________________
dropout_3 (Dropout)          multiple                  0         
_________________________________________________________________
batch_normalization_3 (Batch multiple                  400       
_________________________________________________________________
dense_4 (Dense)              multiple                  10100     
_________________________________________________________________
dropout_4 (Dropout)          multiple                  0         
_________________________________________________________________
batch_normalization_4 (Batch multiple                  400       
_________________________________________________________________
dense_5 (Dense)              multiple                  202       
=================================================================
Total params: 76,362
Trainable params: 75,306
Non-trainable params: 1,056

---

Prediction Code

class_prob=model.predict(new_array_2.T,batch_size=1)
print(class_prob)
classifications=model.predict_classes(new_array_2.T,batch_size=1)
print(classifications)
print(CATEGORIES[classifications[0]])

output

[[0.39456758 0.39456758]]
[0]
Dog

Answer 1

model.predict just returns the computed output of your model given the input, so all the details you mentioned depend on the model's output, for example, activations at the last layer.

That the model outputs probabilities that sum to one is only produced by the softmax activation at the output layer, to me it looks like your last layer has a different activation, probably sigmoid, which will not produce probabilities that sum to one. You should prefer to use softmax for multi-class classification.