Table 11 Comparison of the proposed method with state-of-the-art neural style transfer approaches.
Method | Content preservation (SSIM) | Style fidelity (style loss) | Processing time (s) | Memory usage (MB) | Observations |
|---|---|---|---|---|---|
Gatys et al.2 | 0.80 | 0.92 | 15.2 | 1500 | High style fidelity but computationally intensive due to iterative optimization |
Johnson et al.8 | 0.85 | 0.88 | 2.3 | 800 | Faster with real-time capability but limited flexibility for multiple styles |
Huang and Belongie9 | 0.82 | 0.85 | 1.5 | 500 | Efficient with arbitrary style transfer but moderate style fidelity |
Proposed method | 0.88 | 0.90 | 2.0 | 600 | Superior balance between style and content with improved efficiency |
