{"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/cost-effective-object-detection-active-sample","title":"Cost-effective Object Detection: Active Sample Mining with Switchable Selection Criteria","arxiv_id":"1807.00147","date":"2018-06-30","proceeding":null,"authors":["Keze Wang","Liang Lin","Xiaopeng Yan","Ziliang Chen","Dongyu Zhang","Lei Zhang"],"abstract":"Though quite challenging, leveraging large-scale unlabeled or partially\nlabeled data in learning systems (e.g., model/classifier training) has\nattracted increasing attentions due to its fundamental importance. To address\nthis problem, many active learning (AL) methods have been proposed that employ\nup-to-date detectors to retrieve representative minority samples according to\npredefined confidence or uncertainty thresholds. However, these AL methods\ncause the detectors to ignore the remaining majority samples (i.e., those with\nlow uncertainty or high prediction confidence). In this work, by developing a\nprincipled active sample mining (ASM) framework, we demonstrate that\ncost-effectively mining samples from these unlabeled majority data is key to\ntraining more powerful object detectors while minimizing user effort.\nSpecifically, our ASM framework involves a switchable sample selection\nmechanism for determining whether an unlabeled sample should be manually\nannotated via AL or automatically pseudo-labeled via a novel self-learning\nprocess. The proposed process can be compatible with mini-batch based training\n(i.e., using a batch of unlabeled or partially labeled data as a one-time\ninput) for object detection. In addition, a few samples with low-confidence\npredictions are selected and annotated via AL. Notably, our method is suitable\nfor object categories that are not seen in the unlabeled data during the\nlearning process. Extensive experiments clearly demonstrate that our ASM\nframework can achieve performance comparable to that of alternative methods but\nwith significantly fewer annotations.","url_abs":"http://arxiv.org/abs/1807.00147v3","url_pdf":"http://arxiv.org/pdf/1807.00147v3.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":"cost-effective-object-detection-active-sample","repo_url":"https://github.com/yanxp/ASM","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"caffe2","reach":null}],"tasks":[{"task_slug":"active-learning","task_name":"Active Learning"},{"task_slug":"object-detection","task_name":"Object Detection"},{"task_slug":"self-learning","task_name":"Self-Learning"},{"task_slug":"object-detection-1","task_name":"object-detection"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1807.00147","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}