The purpose of this research was to evaluate the efficacy of safety management on construction sites and to provide a safety evaluation technique based on the digital management system paradigm. An approach based on the depth separable object detection (DSOD) method is suggested to solve the problem of tower crane detection. By dividing the test set's tower crane photographs into groups according to various scenarios and assessing the combined precision and recall, the DSOD algorithm's efficacy in tower crane detection was confirmed. The experimental findings demonstrate that, even at varying shot distances and levels of occlusion, the DSOD algorithm retained a high level of detection precision and recall. For tower crane detection, the DSOD method performed better than other widely used techniques such as SSD and you only look once, version 5. This study offers trustworthy technical assistance for monitoring potentially dangerous sources on building sites, which is crucial for enhancing construction safety management.
Yang et al. (Wed,) studied this question.