Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386075952> ?p ?o ?g. }
- W4386075952 abstract "In the fashion domain, there exists a variety of vision-and-language (V+L) tasks, including cross-modal retrieval, text-guided image retrieval, multi-modal classification, and image captioning. They differ drastically in each individual input/output format and dataset size. It has been common to design a task-specific model and fine-tune it independently from a pre-trained V+l model (e.g., CLIP). This results in parameter inefficiency and inability to exploit inter-task relatedness. To address such issues, we propose a novel FAshion-focused Multi-task Efficient learning method for Vision-and-Language tasks (FAME-ViL) in this work. Compared with existing approaches, FAME-ViL applies a single model for multiple heterogeneous fashion tasks, therefore being much more parameter-efficient. It is enabled by two novel components: (1) a task-versatile architecture with cross-attention adapters and task-specific adapters integrated into a unified V+L model, and (2) a stable and effective multi-task training strategy that supports learning from heterogeneous data and prevents negative transfer. Extensive experiments on four fashion tasks show that our FAME-ViL can save 61.5% of parameters over alternatives, while significantly outperforming the conventional independently trained single-task models. Code is available at https://github.com/BrandonHanx/FAME-ViL." @default.
- W4386075952 created "2023-08-23" @default.
- W4386075952 creator A5006111469 @default.
- W4386075952 creator A5014436524 @default.
- W4386075952 creator A5028643592 @default.
- W4386075952 creator A5036418431 @default.
- W4386075952 creator A5046046128 @default.
- W4386075952 creator A5066716873 @default.
- W4386075952 date "2023-06-01" @default.
- W4386075952 modified "2023-10-17" @default.
- W4386075952 title "FAME-ViL: Multi-Tasking Vision-Language Model for Heterogeneous Fashion Tasks" @default.
- W4386075952 cites W1956340063 @default.
- W4386075952 cites W2016589492 @default.
- W4386075952 cites W2060277733 @default.
- W4386075952 cites W2157331557 @default.
- W4386075952 cites W2194775991 @default.
- W4386075952 cites W2605066040 @default.
- W4386075952 cites W2897195437 @default.
- W4386075952 cites W2905544595 @default.
- W4386075952 cites W2963168538 @default.
- W4386075952 cites W2963540523 @default.
- W4386075952 cites W2970231061 @default.
- W4386075952 cites W2979826702 @default.
- W4386075952 cites W2997591391 @default.
- W4386075952 cites W3026458074 @default.
- W4386075952 cites W3034727271 @default.
- W4386075952 cites W3035485997 @default.
- W4386075952 cites W3143320354 @default.
- W4386075952 cites W3165938948 @default.
- W4386075952 cites W3172514680 @default.
- W4386075952 cites W3173220247 @default.
- W4386075952 cites W3175684172 @default.
- W4386075952 cites W3176909828 @default.
- W4386075952 cites W3184784418 @default.
- W4386075952 cites W3198377975 @default.
- W4386075952 cites W3203247393 @default.
- W4386075952 cites W3206816211 @default.
- W4386075952 cites W4224929299 @default.
- W4386075952 cites W4226528870 @default.
- W4386075952 cites W4281643269 @default.
- W4386075952 cites W4290927857 @default.
- W4386075952 cites W4292828970 @default.
- W4386075952 cites W4312310776 @default.
- W4386075952 cites W4312749754 @default.
- W4386075952 cites W4312825288 @default.
- W4386075952 cites W4312884055 @default.
- W4386075952 cites W4312910992 @default.
- W4386075952 cites W4312940158 @default.
- W4386075952 cites W4385573463 @default.
- W4386075952 doi "https://doi.org/10.1109/cvpr52729.2023.00262" @default.
- W4386075952 hasPublicationYear "2023" @default.
- W4386075952 type Work @default.
- W4386075952 citedByCount "0" @default.
- W4386075952 crossrefType "proceedings-article" @default.
- W4386075952 hasAuthorship W4386075952A5006111469 @default.
- W4386075952 hasAuthorship W4386075952A5014436524 @default.
- W4386075952 hasAuthorship W4386075952A5028643592 @default.
- W4386075952 hasAuthorship W4386075952A5036418431 @default.
- W4386075952 hasAuthorship W4386075952A5046046128 @default.
- W4386075952 hasAuthorship W4386075952A5066716873 @default.
- W4386075952 hasConcept C107457646 @default.
- W4386075952 hasConcept C115961682 @default.
- W4386075952 hasConcept C119857082 @default.
- W4386075952 hasConcept C134306372 @default.
- W4386075952 hasConcept C136197465 @default.
- W4386075952 hasConcept C137293760 @default.
- W4386075952 hasConcept C154945302 @default.
- W4386075952 hasConcept C157657479 @default.
- W4386075952 hasConcept C162324750 @default.
- W4386075952 hasConcept C165696696 @default.
- W4386075952 hasConcept C175154964 @default.
- W4386075952 hasConcept C177264268 @default.
- W4386075952 hasConcept C185592680 @default.
- W4386075952 hasConcept C187736073 @default.
- W4386075952 hasConcept C188027245 @default.
- W4386075952 hasConcept C199360897 @default.
- W4386075952 hasConcept C2776760102 @default.
- W4386075952 hasConcept C2780451532 @default.
- W4386075952 hasConcept C28006648 @default.
- W4386075952 hasConcept C33923547 @default.
- W4386075952 hasConcept C36503486 @default.
- W4386075952 hasConcept C38652104 @default.
- W4386075952 hasConcept C41008148 @default.
- W4386075952 hasConcept C71139939 @default.
- W4386075952 hasConceptScore W4386075952C107457646 @default.
- W4386075952 hasConceptScore W4386075952C115961682 @default.
- W4386075952 hasConceptScore W4386075952C119857082 @default.
- W4386075952 hasConceptScore W4386075952C134306372 @default.
- W4386075952 hasConceptScore W4386075952C136197465 @default.
- W4386075952 hasConceptScore W4386075952C137293760 @default.
- W4386075952 hasConceptScore W4386075952C154945302 @default.
- W4386075952 hasConceptScore W4386075952C157657479 @default.
- W4386075952 hasConceptScore W4386075952C162324750 @default.
- W4386075952 hasConceptScore W4386075952C165696696 @default.
- W4386075952 hasConceptScore W4386075952C175154964 @default.
- W4386075952 hasConceptScore W4386075952C177264268 @default.
- W4386075952 hasConceptScore W4386075952C185592680 @default.
- W4386075952 hasConceptScore W4386075952C187736073 @default.
- W4386075952 hasConceptScore W4386075952C188027245 @default.
- W4386075952 hasConceptScore W4386075952C199360897 @default.