{"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/hollywood-in-homes-crowdsourcing-data","title":"Hollywood in Homes: Crowdsourcing Data Collection for Activity Understanding","arxiv_id":"1604.01753","date":"2016-04-06","proceeding":null,"authors":["Gunnar A. Sigurdsson","Gül Varol","Xiaolong Wang","Ali Farhadi","Ivan Laptev","Abhinav Gupta"],"abstract":"Computer vision has a great potential to help our daily lives by searching\nfor lost keys, watering flowers or reminding us to take a pill. To succeed with\nsuch tasks, computer vision methods need to be trained from real and diverse\nexamples of our daily dynamic scenes. While most of such scenes are not\nparticularly exciting, they typically do not appear on YouTube, in movies or TV\nbroadcasts. So how do we collect sufficiently many diverse but boring samples\nrepresenting our lives? We propose a novel Hollywood in Homes approach to\ncollect such data. Instead of shooting videos in the lab, we ensure diversity\nby distributing and crowdsourcing the whole process of video creation from\nscript writing to video recording and annotation. Following this procedure we\ncollect a new dataset, Charades, with hundreds of people recording videos in\ntheir own homes, acting out casual everyday activities. The dataset is composed\nof 9,848 annotated videos with an average length of 30 seconds, showing\nactivities of 267 people from three continents. Each video is annotated by\nmultiple free-text descriptions, action labels, action intervals and classes of\ninteracted objects. In total, Charades provides 27,847 video descriptions,\n66,500 temporally localized intervals for 157 action classes and 41,104 labels\nfor 46 object classes. Using this rich data, we evaluate and provide baseline\nresults for several tasks including action recognition and automatic\ndescription generation. We believe that the realism, diversity, and casual\nnature of this dataset will present unique challenges and new opportunities for\ncomputer vision community.","url_abs":"http://arxiv.org/abs/1604.01753v3","url_pdf":"http://arxiv.org/pdf/1604.01753v3.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":"action-recognition-in-videos","task_name":"Action Recognition"},{"task_slug":"diversity","task_name":"Diversity"},{"task_slug":"action-recognition","task_name":"Temporal Action Localization"}],"methods":[],"datasets_introduced":[{"slug":"charades","name":"Charades","full_name":""}],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1604.01753","atlas_url":"https://app.syntology.ai/?focus=1604.01753","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}