{"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/fixed-sized-representation-learning-from","title":"Fixed-sized representation learning from Offline Handwritten Signatures of different sizes","arxiv_id":"1804.00448","date":"2018-04-02","proceeding":null,"authors":["Luiz G. Hafemann","Robert Sabourin","Luiz S. Oliveira"],"abstract":"Methods for learning feature representations for Offline Handwritten\nSignature Verification have been successfully proposed in recent literature,\nusing Deep Convolutional Neural Networks to learn representations from\nsignature pixels. Such methods reported large performance improvements compared\nto handcrafted feature extractors. However, they also introduced an important\nconstraint: the inputs to the neural networks must have a fixed size, while\nsignatures vary significantly in size between different users. In this paper we\npropose addressing this issue by learning a fixed-sized representation from\nvariable-sized signatures by modifying the network architecture, using Spatial\nPyramid Pooling. We also investigate the impact of the resolution of the images\nused for training, and the impact of adapting (fine-tuning) the representations\nto new operating conditions (different acquisition protocols, such as writing\ninstruments and scan resolution). On the GPDS dataset, we achieve results\ncomparable with the state-of-the-art, while removing the constraint of having a\nmaximum size for the signatures to be processed. We also show that using higher\nresolutions (300 or 600dpi) can improve performance when skilled forgeries from\na subset of users are available for feature learning, but lower resolutions\n(around 100dpi) can be used if only genuine signatures are used. Lastly, we\nshow that fine-tuning can improve performance when the operating conditions\nchange.","url_abs":"http://arxiv.org/abs/1804.00448v2","url_pdf":"http://arxiv.org/pdf/1804.00448v2.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":"fixed-sized-representation-learning-from","repo_url":"https://github.com/luizgh/sigver_wiwd","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"representation-learning","task_name":"Representation Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}