HyperLoRA: Parameter-Efficient Adaptive Generation for Portrait Synthesis
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866913749815787520 |
|---|---|
| author | Li, Mengtian Chen, Jinshu Feng, Wanquan Li, Bingchuan Dai, Fei Zhao, Songtao He, Qian |
| author_facet | Li, Mengtian Chen, Jinshu Feng, Wanquan Li, Bingchuan Dai, Fei Zhao, Songtao He, Qian |
| contents | Personalized portrait synthesis, essential in domains like social entertainment, has recently made significant progress. Person-wise fine-tuning based methods, such as LoRA and DreamBooth, can produce photorealistic outputs but need training on individual samples, consuming time and resources and posing an unstable risk. Adapter based techniques such as IP-Adapter freeze the foundational model parameters and employ a plug-in architecture to enable zero-shot inference, but they often exhibit a lack of naturalness and authenticity, which are not to be overlooked in portrait synthesis tasks. In this paper, we introduce a parameter-efficient adaptive generation method, namely HyperLoRA, that uses an adaptive plug-in network to generate LoRA weights, merging the superior performance of LoRA with the zero-shot capability of adapter scheme. Through our carefully designed network structure and training strategy, we achieve zero-shot personalized portrait generation (supporting both single and multiple image inputs) with high photorealism, fidelity, and editability. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_16944 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | HyperLoRA: Parameter-Efficient Adaptive Generation for Portrait Synthesis Li, Mengtian Chen, Jinshu Feng, Wanquan Li, Bingchuan Dai, Fei Zhao, Songtao He, Qian Computer Vision and Pattern Recognition Personalized portrait synthesis, essential in domains like social entertainment, has recently made significant progress. Person-wise fine-tuning based methods, such as LoRA and DreamBooth, can produce photorealistic outputs but need training on individual samples, consuming time and resources and posing an unstable risk. Adapter based techniques such as IP-Adapter freeze the foundational model parameters and employ a plug-in architecture to enable zero-shot inference, but they often exhibit a lack of naturalness and authenticity, which are not to be overlooked in portrait synthesis tasks. In this paper, we introduce a parameter-efficient adaptive generation method, namely HyperLoRA, that uses an adaptive plug-in network to generate LoRA weights, merging the superior performance of LoRA with the zero-shot capability of adapter scheme. Through our carefully designed network structure and training strategy, we achieve zero-shot personalized portrait generation (supporting both single and multiple image inputs) with high photorealism, fidelity, and editability. |
| title | HyperLoRA: Parameter-Efficient Adaptive Generation for Portrait Synthesis |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2503.16944 |