Dimensions overall score 7.0

• MeInTime extends reference-based face restoration from same-age to cross-age settings.method 95%
• MeInTime decouples the modeling of identity and age conditions during training.technical 90%
• MeInTime introduces a new attention mechanism to effectively inject identity features during training.technical 90%
• MeInTime uses Gated Residual Fusion modules to integrate degraded features with identity representations.technical 90%
• MeInTime proposes Age-Aware Gradient Guidance, a training-free sampling strategy that uses age-driven direction to guide denoising.method 90%
• MeInTime outperforms existing face restoration methods in both identity preservation and age consistency.result 85%
• Most existing reference-based face restoration methods assume reference and degraded input are age-aligned, limiting effectiveness in real-world cross-age scenarios.limitation 85%
• MeInTime can work with one or few reference images along with an age prompt to achieve faithful restoration.method 85%

Competitive landscape

Competitor map is still being generated for this paper. Enable generation or check back soon.

Startup potential card