The research focuses on developing No-Reference Image Quality Assessment (NR-IQA) methods using deep learning techniques to evaluate the perceptual quality of images without requiring a reference image. By leveraging advanced neural networks, including CNNs and Transformers, the study aims to extract meaningful features that align with the Human Visual System (HVS), ensuring accurate quality predictions. This approach is particularly valuable for applications in multimedia, medical imaging, and emerging technologies like AR/VR, where reference images are often unavailable.