{"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/recognizing-fine-grained-and-composite","title":"Recognizing Fine-Grained and Composite Activities using Hand-Centric Features and Script Data","arxiv_id":"1502.06648","date":"2015-02-23","proceeding":null,"authors":["Marcus Rohrbach","Anna Rohrbach","Michaela Regneri","Sikandar Amin","Mykhaylo Andriluka","Manfred Pinkal","Bernt Schiele"],"abstract":"Activity recognition has shown impressive progress in recent years. However,\nthe challenges of detecting fine-grained activities and understanding how they\nare combined into composite activities have been largely overlooked. In this\nwork we approach both tasks and present a dataset which provides detailed\nannotations to address them. The first challenge is to detect fine-grained\nactivities, which are defined by low inter-class variability and are typically\ncharacterized by fine-grained body motions. We explore how human pose and hands\ncan help to approach this challenge by comparing two pose-based and two\nhand-centric features with state-of-the-art holistic features. To attack the\nsecond challenge, recognizing composite activities, we leverage the fact that\nthese activities are compositional and that the essential components of the\nactivities can be obtained from textual descriptions or scripts. We show the\nbenefits of our hand-centric approach for fine-grained activity classification\nand detection. For composite activity recognition we find that decomposition\ninto attributes allows sharing information across composites and is essential\nto attack this hard task. Using script data we can recognize novel composites\nwithout having training data for them.","url_abs":"http://arxiv.org/abs/1502.06648v2","url_pdf":"http://arxiv.org/pdf/1502.06648v2.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":[],"tasks":[{"task_slug":"activity-recognition","task_name":"Activity Recognition"}],"methods":[],"datasets_introduced":[{"slug":"mpii-cooking-2-dataset","name":"MPII Cooking 2 Dataset","full_name":""}],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1502.06648","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}