SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2306.08789 · 2023

Efficient Token-Guided Image-Text Retrieval with Consistent Multimodal Contrastive Training

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 15 functions out of this paper's own repositories and ran 10 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
lcfractal/tgdt pwc_unofficial 10 of 15
FunctionStatusWhere it lives
EncoderText Ran lcfractal/tgdt/models/text.py
code served (permissive licence) · get_code("e1d7360c5be25e25")
cosine_sim Ran lcfractal/tgdt/utils.py
code served (permissive licence) · get_code("f2e17288ada8cf89")
dot_sim Ran lcfractal/tgdt/models/loss.py
code served (permissive licence) · get_code("6a49bab714ec6122")
dot_sim Ran lcfractal/tgdt/utils.py
code served (permissive licence) · get_code("36545e75ca862270")
find_nhead Ran lcfractal/tgdt/models/visual.py
code served (permissive licence) · get_code("6e963f0bd8b4ec0f")
generate_square_subsequent_mask Ran lcfractal/tgdt/models/utils.py
code served (permissive licence) · get_code("894b253f35765276")
get_paths Ran lcfractal/tgdt/data.py
code served (permissive licence) · get_code("66f891f51e3bec44")
get_transform Ran lcfractal/tgdt/data.py
code served (permissive licence) · get_code("e7518699cbc1bd68")
l2norm Ran lcfractal/tgdt/models/utils.py
code served (permissive licence) · get_code("f92e514b398e75d8")
my_collate Ran lcfractal/tgdt/evaluate_utils/compute_relevance.py
code served (permissive licence) · get_code("d9047caaedf4aadb")
accuracy Not yet run lcfractal/tgdt/train_ft.py
code served (permissive licence) · get_code("f0c9a29156911331")
cosine_sim Not yet run lcfractal/tgdt/models/loss.py
code served (permissive licence) · get_code("7f3908ad08980104")
encode_data Not yet run lcfractal/tgdt/evaluation.py
code served (permissive licence) · get_code("5a365ec8a05fffb9")
get_model Not yet run lcfractal/tgdt/utils.py
code served (permissive licence) · get_code("a1ea9bbed7fcc094")
order_sim Not yet run lcfractal/tgdt/models/loss.py
code served (permissive licence) · get_code("da16023b8239c604")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Image-text retrieval is a central problem for understanding the semantic relationship between vision and language, and serves as the basis for various visual and language tasks. Most previous works either simply learn coarse-grained representations of the overall image and text, or elaborately establish the correspondence between image regions or pixels and text words. However, the close relations between coarse- and fine-grained representations for each modality are important for image-text retrieval but almost neglected. As a result, such previous works inevitably suffer from low retrieval accuracy or heavy computational cost. In this work, we address image-text retrieval from a novel perspective by combining coarse- and fine-grained representation learning into a unified framework. This framework is consistent with human cognition, as humans simultaneously pay attention to the entire sample and regional elements to understand the semantic content. To this end, a Token-Guided Dual Transformer (TGDT) architecture which consists of two homogeneous branches for image and text modalities, respectively, is proposed for image-text retrieval. The TGDT incorporates both coarse- and fine-grained retrievals into a unified framework and beneficially leverages the advantages of both retrieval approaches. A novel training objective called Consistent Multimodal Contrastive (CMC) loss is proposed accordingly to ensure the intra- and inter-modal semantic consistencies between images and texts in the common embedding space. Equipped with a two-stage inference method based on the mixed global and local cross-modal similarity, the proposed method achieves state-of-the-art retrieval performances with extremely low inference time when compared with representative recent approaches.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2306.08789")
get_code_for_paper("2306.08789")
have("2306.08789")

Connect an agent — have() is free.