Table 3
Multi-objective optimization performance metric definitions and measurement methods.
| Indicator name | Calculation method | Optimization direction | Objective weight |
|---|---|---|---|
| Style quality (FID) | Fréchet distance between generated image and target style in feature space | The smaller the better | 0.28 |
| Interaction latency (ms) | End-to-end time from user action event to interface rendering completion | The smaller the better | 0.23 |
| User satisfaction (%) | The weighted inverse of the normalized task completion time and the number of steps is used | Bigger is better | 0.19 |
| Design diversity | LPIPS pairwise mean of 10 generated results under the same content input | Bigger is better | 0.15 |
| Constraint violation rate (%) | (WCAG violations + layout violations) / total inspection items × 100% | Bigger is better | 0.15 |
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.