This paper presents a benchmark for salient object detection and segmentation, comparing 41 state-of-the-art models across seven challenging datasets. The results show significant progress in both accuracy and efficiency over the past few years. The top models outperform previous benchmarks, with salient object detection models generally performing better than related models. The study analyzes the impact of center bias and scene complexity on model performance, and proposes solutions to address open problems such as evaluation scores and dataset bias. The benchmark includes models for salient object detection, fixation prediction, and object proposal generation, with a baseline model to study center bias. The datasets used are diverse, covering different biases and complexities. Evaluation metrics include precision-recall, F-measure, ROC, AUC, MAE, and Fβw-measure. The results show that DRFI and DSR models perform best, while some models struggle with center bias and complex scenes. The study also highlights the importance of considering background prior and the need for more challenging datasets. The paper concludes that salient object detection remains an active research area with potential applications in various fields. The benchmark provides a comprehensive evaluation of models and highlights the challenges in this field.This paper presents a benchmark for salient object detection and segmentation, comparing 41 state-of-the-art models across seven challenging datasets. The results show significant progress in both accuracy and efficiency over the past few years. The top models outperform previous benchmarks, with salient object detection models generally performing better than related models. The study analyzes the impact of center bias and scene complexity on model performance, and proposes solutions to address open problems such as evaluation scores and dataset bias. The benchmark includes models for salient object detection, fixation prediction, and object proposal generation, with a baseline model to study center bias. The datasets used are diverse, covering different biases and complexities. Evaluation metrics include precision-recall, F-measure, ROC, AUC, MAE, and Fβw-measure. The results show that DRFI and DSR models perform best, while some models struggle with center bias and complex scenes. The study also highlights the importance of considering background prior and the need for more challenging datasets. The paper concludes that salient object detection remains an active research area with potential applications in various fields. The benchmark provides a comprehensive evaluation of models and highlights the challenges in this field.