AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

Cross-Domain Image Captioning With Discriminative Finetuning

Roberto Dessì, Michele Bevilacqua, Eleonora Gualdoni, Nathanaël Carraz Rakotonirina, Francesca Franzon, Marco Baroni

CVPR 2023 · Vision-language and multimodal › Human and embodied modeling · approach: Neural image captioning · 21 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Fine-tune ClipCap with a discriminative retrieval objectiveClipCap-COCO · COCOP@1: 74.2 -> 84.8Improved
Fine-tune ClipCap with a discriminative retrieval objectiveClipCap-ConCap · Conceptual CaptionsP@1: 82.5 -> 94.4Improved
Fine-tune ClipCap with a discriminative retrieval objectiveClipCap-ConCap · FlickrP@1: 76.7 -> 87.8Improved
Fine-tune ClipCap with a discriminative retrieval objectiveClipCap-ConCap · ConcadiaP@1: 59.17 -> 80.49Improved
Fine-tune ClipCap with a discriminative retrieval objectiveClipCap-COCO · COCOCIDEr: 108.55 -> 105.40Hurt
Fine-tune ClipCap with a discriminative retrieval objectiveClipCap-ConCap · Conceptual CaptionsCIDEr: 87.22 -> 55.26Hurt
Fine-tune BLIP with a discriminative retrieval objectiveBLIP · FlickrP@1: 75.1 -> 80.8Improved
Use discriminatively fine-tuned captions instead of human captionsConceptual Captions human image identification · Conceptual Captions human retrievalaccuracy: 42.8 -> 47.6Improved
Remove the REINFORCE baseline (ablation)ClipCap trained on Flickr · Flickr validationP@1: 97.8 -> 34.3Hurt
Replace the running mean with greedy decoding as the REINFORCE baseline (ablation)ClipCap trained on Flickr · Flickr validationP@1: 97.8 vs 97.4No clear effect
Replace log-probability retrieval reward with cosine similarity reward (ablation)ClipCap trained on Flickr · Flickr validationP@1: 86.2 -> 85.2Hurt
Replace log-probability retrieval reward with accuracy reward (ablation)ClipCap trained on Flickr · Flickr validationP@1: 86.2 -> 85.3Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy