Matches in SemOpenAlex for { <https://semopenalex.org/work/W4384659783> ?p ?o ?g. }
Showing items 1 to 76 of
76
with 100 items per page.
- W4384659783 abstract "Vision-language (VL) Pre-training (VLP) has shown to well generalize VL models over a wide range of VL downstream tasks, especially for cross-modal retrieval. However, it hinges on a huge amount of image-text pairs, which requires tedious and costly curation. On the contrary, weakly-supervised VLP (W-VLP) explores means with object tags generated by a pre-trained object detector (OD) from images. Yet, they still require paired information, i.e. images and object-level annotations, as supervision to train an OD. To further reduce the amount of supervision, we propose Prompts-in-The-Loop (PiTL) that prompts knowledge from large language models (LLMs) to describe images. Concretely, given a category label of an image, e.g. refinery, the knowledge, e.g. a refinery could be seen with large storage tanks, pipework, and ..., extracted by LLMs is used as the language counterpart. The knowledge supplements, e.g. the common relations among entities most likely appearing in a scene. We create IN14K, a new VL dataset of 9M images and 1M descriptions of 14K categories from ImageNet21K with PiTL. Empirically, the VL models pre-trained with PiTL-generated pairs are strongly favored over other W-VLP works on image-to-text (I2T) and text-to-image (T2I) retrieval tasks, with less supervision. The results reveal the effectiveness of PiTL-generated pairs for VLP." @default.
- W4384659783 created "2023-07-20" @default.
- W4384659783 creator A5036133390 @default.
- W4384659783 creator A5049343215 @default.
- W4384659783 creator A5071798070 @default.
- W4384659783 creator A5077052515 @default.
- W4384659783 creator A5079044890 @default.
- W4384659783 date "2023-07-18" @default.
- W4384659783 modified "2023-10-14" @default.
- W4384659783 title "PiTL: Cross-modal Retrieval with Weakly-supervised Vision-language Pre-training via Prompting" @default.
- W4384659783 cites W1566289585 @default.
- W4384659783 cites W2108598243 @default.
- W4384659783 cites W2886641317 @default.
- W4384659783 cites W3090449556 @default.
- W4384659783 cites W3173220247 @default.
- W4384659783 cites W3176641147 @default.
- W4384659783 cites W3177224328 @default.
- W4384659783 cites W4285606530 @default.
- W4384659783 cites W4312877428 @default.
- W4384659783 cites W4312910992 @default.
- W4384659783 cites W4385573407 @default.
- W4384659783 cites W4385574358 @default.
- W4384659783 doi "https://doi.org/10.1145/3539618.3592038" @default.
- W4384659783 hasPublicationYear "2023" @default.
- W4384659783 type Work @default.
- W4384659783 citedByCount "0" @default.
- W4384659783 crossrefType "proceedings-article" @default.
- W4384659783 hasAuthorship W4384659783A5036133390 @default.
- W4384659783 hasAuthorship W4384659783A5049343215 @default.
- W4384659783 hasAuthorship W4384659783A5071798070 @default.
- W4384659783 hasAuthorship W4384659783A5077052515 @default.
- W4384659783 hasAuthorship W4384659783A5079044890 @default.
- W4384659783 hasBestOaLocation W43846597831 @default.
- W4384659783 hasConcept C115961682 @default.
- W4384659783 hasConcept C153180895 @default.
- W4384659783 hasConcept C154945302 @default.
- W4384659783 hasConcept C1667742 @default.
- W4384659783 hasConcept C185592680 @default.
- W4384659783 hasConcept C188027245 @default.
- W4384659783 hasConcept C204321447 @default.
- W4384659783 hasConcept C23123220 @default.
- W4384659783 hasConcept C2776151529 @default.
- W4384659783 hasConcept C2781238097 @default.
- W4384659783 hasConcept C31972630 @default.
- W4384659783 hasConcept C41008148 @default.
- W4384659783 hasConcept C71139939 @default.
- W4384659783 hasConceptScore W4384659783C115961682 @default.
- W4384659783 hasConceptScore W4384659783C153180895 @default.
- W4384659783 hasConceptScore W4384659783C154945302 @default.
- W4384659783 hasConceptScore W4384659783C1667742 @default.
- W4384659783 hasConceptScore W4384659783C185592680 @default.
- W4384659783 hasConceptScore W4384659783C188027245 @default.
- W4384659783 hasConceptScore W4384659783C204321447 @default.
- W4384659783 hasConceptScore W4384659783C23123220 @default.
- W4384659783 hasConceptScore W4384659783C2776151529 @default.
- W4384659783 hasConceptScore W4384659783C2781238097 @default.
- W4384659783 hasConceptScore W4384659783C31972630 @default.
- W4384659783 hasConceptScore W4384659783C41008148 @default.
- W4384659783 hasConceptScore W4384659783C71139939 @default.
- W4384659783 hasLocation W43846597831 @default.
- W4384659783 hasLocation W43846597832 @default.
- W4384659783 hasOpenAccess W4384659783 @default.
- W4384659783 hasPrimaryLocation W43846597831 @default.
- W4384659783 hasRelatedWork W1837097281 @default.
- W4384659783 hasRelatedWork W1966410754 @default.
- W4384659783 hasRelatedWork W1988485990 @default.
- W4384659783 hasRelatedWork W2007544051 @default.
- W4384659783 hasRelatedWork W2095705906 @default.
- W4384659783 hasRelatedWork W2334336442 @default.
- W4384659783 hasRelatedWork W2732308154 @default.
- W4384659783 hasRelatedWork W2922421953 @default.
- W4384659783 hasRelatedWork W2975200075 @default.
- W4384659783 hasRelatedWork W3177406559 @default.
- W4384659783 isParatext "false" @default.
- W4384659783 isRetracted "false" @default.
- W4384659783 workType "article" @default.