With its myriad applications in pattern recognition, the identification of salient objects presents a significant challenge to computer vision. Despite the existence of numerous proposed models, there remains ample scope for further research in this field. This survey work provides a comprehensive overview of recent advancements in saliency detection, encompassing both heuristic- and deep learning-based methodologies. The survey explores various related domains, including models for video saliency detection and prediction of eye fixations. Additionally, it investigates RGBD salient object recognition and co-saliency object detection. A critical analysis of algorithm taxonomy, unresolved challenges, network design, supervision levels, learning paradigms, and evaluation metrics is presented. Furthermore, this study offers valuable insights into future trends and addresses significant concerns within saliency models, particularly focusing on deep learning-based solutions. Benchmarking of representative SOD (Salient Object Detection) models is conducted, accompanied by comparative analyses, investigations into robustness against perturbations and adversarial attacks, discussions on dataset generalization and difficulty, and the introduction of a novel SOD dataset enriched with attribute annotations. In summary, this survey work serves as a comprehensive resource for researchers and practitioners in the field of computer vision, providing an in-depth examination of recent developments, addressing key challenges, and offering guidance for future research directions.