{"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/spp-net-deep-absolute-pose-regression-with","title":"SPP-Net: Deep Absolute Pose Regression with Synthetic Views","arxiv_id":"1712.03452","date":"2017-12-09","proceeding":null,"authors":["Pulak Purkait","Cheng Zhao","Christopher Zach"],"abstract":"Image based localization is one of the important problems in computer vision\ndue to its wide applicability in robotics, augmented reality, and autonomous\nsystems. There is a rich set of methods described in the literature how to\ngeometrically register a 2D image w.r.t.\\ a 3D model. Recently, methods based\non deep (and convolutional) feedforward networks (CNNs) became popular for pose\nregression. However, these CNN-based methods are still less accurate than\ngeometry based methods despite being fast and memory efficient. In this work we\ndesign a deep neural network architecture based on sparse feature descriptors\nto estimate the absolute pose of an image. Our choice of using sparse feature\ndescriptors has two major advantages: first, our network is significantly\nsmaller than the CNNs proposed in the literature for this task---thereby making\nour approach more efficient and scalable. Second---and more importantly---,\nusage of sparse features allows to augment the training data with synthetic\nviewpoints, which leads to substantial improvements in the generalization\nperformance to unseen poses. Thus, our proposed method aims to combine the best\nof the two worlds---feature-based localization and CNN-based pose\nregression--to achieve state-of-the-art performance in the absolute pose\nestimation. A detailed analysis of the proposed architecture and a rigorous\nevaluation on the existing datasets are provided to support our method.","url_abs":"http://arxiv.org/abs/1712.03452v1","url_pdf":"http://arxiv.org/pdf/1712.03452v1.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":"spp-net-deep-absolute-pose-regression-with","repo_url":"https://github.com/Mind23-2/MindCode-173","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":null}],"tasks":[{"task_slug":"image-based-localization","task_name":"Image-Based Localization"},{"task_slug":"pose-estimation","task_name":"Pose Estimation"},{"task_slug":"regression-1","task_name":"regression"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}