Overfitting happens when the model fits too well to the training set. It then becomes difficult for the model to generalize to new examples that were not in the training set. For example, model recognizes specific images in the training set instead of general patterns. Training accuracy will be higher than the accuracy on the validation/test set.
Steps for reducing overfitting:
Add more variant Dataset
Make use of balance Dataset
Use data augmentation
Use architectures that generalize well
Add regularization (mostly dropout, L1/L2 regularization are also possible)
Refer to the following link for further information: