Have you tried introducing artificial variations to expand the dataset, enhancing model generalization? You can also explore constraining model complexity to prevent overfitting and promote better generalization.
"The use of data augmentation, adjusting the learning rate, reducing model complexity, adjusting the batch size, utilizing regularization techniques, testing various optimizers, appropriately initializing the weights, and adjusting the hyperparameters can all be used to address constant validation accuracy in the CNN"