{"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/towards-dense-object-tracking-in-a-2d","title":"Towards dense object tracking in a 2D honeybee hive","arxiv_id":"1712.08324","date":"2017-12-22","proceeding":"CVPR 2018 6","authors":["Katarzyna Bozek","Laetitia Hebert","Alexander S Mikheyev","Greg J. Stephens"],"abstract":"From human crowds to cells in tissue, the detection and efficient tracking of\nmultiple objects in dense configurations is an important and unsolved problem.\nIn the past, limitations of image analysis have restricted studies of dense\ngroups to tracking a single or subset of marked individuals, or to\ncoarse-grained group-level dynamics, all of which yield incomplete information.\nHere, we combine convolutional neural networks (CNNs) with the model\nenvironment of a honeybee hive to automatically recognize all individuals in a\ndense group from raw image data. We create new, adapted individual labeling and\nuse the segmentation architecture U-Net with a loss function dependent on both\nobject identity and orientation. We additionally exploit temporal regularities\nof the video recording in a recurrent manner and achieve near human-level\nperformance while reducing the network size by 94% compared to the original\nU-Net architecture. Given our novel application of CNNs, we generate extensive\nproblem-specific image data in which labeled examples are produced through a\ncustom interface with Amazon Mechanical Turk. This dataset contains over\n375,000 labeled bee instances across 720 video frames at 2 FPS, representing an\nextensive resource for the development and testing of tracking methods. We\ncorrectly detect 96% of individuals with a location error of ~7% of a typical\nbody dimension, and orientation error of 12 degrees, approximating the\nvariability of human raters. Our results provide an important step towards\nefficient image-based dense object tracking by allowing for the accurate\ndetermination of object location and orientation across time-series image data\nefficiently within one network architecture.","url_abs":"http://arxiv.org/abs/1712.08324v1","url_pdf":"http://arxiv.org/pdf/1712.08324v1.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":"towards-dense-object-tracking-in-a-2d","repo_url":"https://github.com/oist/DenseObjectAnnotation","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"none","reach":{"status":"unanswered"}},{"paper_slug":"towards-dense-object-tracking-in-a-2d","repo_url":"https://github.com/oist/DenseObjectDetection","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"tf","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"object","task_name":"Object"},{"task_slug":"object-detection","task_name":"Object Detection"},{"task_slug":"object-tracking","task_name":"Object Tracking"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"},{"task_slug":"time-series","task_name":"Time Series Analysis"}],"methods":[{"method_slug":"concatenated-skip-connection","method_name":"Concatenated Skip Connection"},{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"max-pooling","method_name":"Max Pooling"},{"method_slug":"relu","method_name":"ReLU"},{"method_slug":"u-net","method_name":"U-Net"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1712.08324","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}