Text this: Classification of cervical cancer using Dense CapsNet with Seg-UNet and denoising autoencoders