English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Curation methods for massive vision-language datasets trade off betweendataset size and quality. However, even the highest quality of availablecurated captions are far too short to capture the rich visual detail in animage. To show the value of dense and highly-aligned image-text pairs, wecollect the Densely Captioned Images (DCI) dataset, containing 8012 naturalimages human-annotated with mask-aligned descriptions averaging above 1000words each. With precise and reliable captions associated with specific partsof an image, we can evaluate vision-language models' (VLMs) understanding ofimage content with a novel task that matches each caption with itscorresponding subcrop. As current models are often limited to 77 text tokens,we also introduce a summarized version (sDCI) in which each caption length islimited. We show that modern techniques that make progress on standardbenchmarks do not correspond with significant improvement on our sDCI basedbenchmark. Lastly, we finetune CLIP using sDCI and show significantimprovements over the baseline despite a small training set. By releasing thefirst human annotated dense image captioning dataset, we hope to enable thedevelopment of new benchmarks or fine-tuning recipes for the next generation ofVLMs to come.

A Picture is Worth More Than 77 Text Tokens: Evaluating CLIP-Style Models on Dense Captions