{"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/dag-recurrent-neural-networks-for-scene","title":"DAG-Recurrent Neural Networks For Scene Labeling","arxiv_id":"1509.00552","date":"2015-09-02","proceeding":"CVPR 2016 6","authors":["Bing Shuai","Zhen Zuo","Gang Wang","Bing Wang"],"abstract":"In image labeling, local representations for image units are usually\ngenerated from their surrounding image patches, thus long-range contextual\ninformation is not effectively encoded. In this paper, we introduce recurrent\nneural networks (RNNs) to address this issue. Specifically, directed acyclic\ngraph RNNs (DAG-RNNs) are proposed to process DAG-structured images, which\nenables the network to model long-range semantic dependencies among image\nunits. Our DAG-RNNs are capable of tremendously enhancing the discriminative\npower of local representations, which significantly benefits the local\nclassification. Meanwhile, we propose a novel class weighting function that\nattends to rare classes, which phenomenally boosts the recognition accuracy for\nnon-frequent classes. Integrating with convolution and deconvolution layers,\nour DAG-RNNs achieve new state-of-the-art results on the challenging SiftFlow,\nCamVid and Barcelona benchmarks.","url_abs":"http://arxiv.org/abs/1509.00552v2","url_pdf":"http://arxiv.org/pdf/1509.00552v2.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":"classification","task_name":"General Classification"},{"task_slug":"scene-labeling","task_name":"Scene Labeling"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"}],"methods":[{"method_slug":"convolution","method_name":"Convolution"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/semantic-segmentation-on-coco-stuff-test","task":"Semantic Segmentation","dataset":"COCO-Stuff test","model":"DAG-RNN (VGG-16)","rank_in_archive_order":20,"of":21,"metrics":{"mIoU":"31.2%"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/1509.00552","atlas_url":"https://app.syntology.ai/?focus=1509.00552","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}