Table 6 Comparative Optimization Results.

From: Real-time construction safety monitoring using a drone based deep hybrid attention model

Optimizer

Best Accuracy (%)

Mean Accuracy (%)

Std. Dev

Final Loss

Convergence Behavior

SGD

96.45

95.87

0.48

0.095

Slow convergence

Adam

97.82

97.43

0.33

0.071

Fast but oscillatory

PSO

98.11

97.90

0.27

0.064

Moderate stability

OOA (Proposed)

99.12

99.05

0.11

0.031

Smooth, stable convergence

  1. Indicate the best-performing results for each evaluation metric among all comparedmethods or configurations. For metrics where higher values indicate better performance, the highest value is highlighted in bold; forerror- or deviation-based metrics, the lowest value is highlighted in bold.
Back to article page