{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/visual-referring-expression-recognition-what","title":"Visual Referring Expression Recognition: What Do Systems Actually Learn?","arxiv_id":"1805.11818","date":"2018-05-30","proceeding":"NAACL 2018 6","authors":["Volkan Cirik","Louis-Philippe Morency","Taylor Berg-Kirkpatrick"],"abstract":"We present an empirical analysis of the state-of-the-art systems for\nreferring expression recognition -- the task of identifying the object in an\nimage referred to by a natural language expression -- with the goal of gaining\ninsight into how these systems reason about language and vision. Surprisingly,\nwe find strong evidence that even sophisticated and linguistically-motivated\nmodels for this task may ignore the linguistic structure, instead relying on\nshallow correlations introduced by unintended biases in the data selection and\nannotation process. For example, we show that a system trained and tested on\nthe input image $\\textit{without the input referring expression}$ can achieve a\nprecision of 71.2% in top-2 predictions. Furthermore, a system that predicts\nonly the object category given the input can achieve a precision of 84.2% in\ntop-2 predictions. These surprisingly positive results for what should be\ndeficient prediction scenarios suggest that careful analysis of what our models\nare learning -- and further, how our data is constructed -- is critical as we\nseek to make substantive progress on grounded language tasks.","url_abs":"http://arxiv.org/abs/1805.11818v1","url_pdf":"http://arxiv.org/pdf/1805.11818v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"visual-referring-expression-recognition-what","repo_url":"https://github.com/volkancirik/neural-sieves-refexp","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"referring-expression","task_name":"Referring Expression"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1805.11818","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}