Resnet101 Model Transfer learning issue

Hi. I am trying to train Resnet101 with insect species dataset. It shows validation accuracy 99% and Validation Loss 28%. But when I test this model after training, to predict an image it always show wrong prediction.
Same pre-trained model I had used in MATLAB it showed same accuracy and loss. but when I Predicted it show 90% correct prediction, while in TensorFlow it show always wrong prediction. what is issue could you please guide me?

Hi @haleema_sadia, As your validation loss is 28% which is on the higher side so the model was not able to predict it properly. You have to minimize the loss as low as possible to get the accurate predictions. Thank You.