{"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/understanding-the-limitations-of-cnn-based","title":"Understanding the Limitations of CNN-based Absolute Camera Pose Regression","arxiv_id":"1903.07504","date":"2019-03-18","proceeding":"CVPR 2019 6","authors":["Torsten Sattler","Qunjie Zhou","Marc Pollefeys","Laura Leal-Taixe"],"abstract":"Visual localization is the task of accurate camera pose estimation in a known\nscene. It is a key problem in computer vision and robotics, with applications\nincluding self-driving cars, Structure-from-Motion, SLAM, and Mixed Reality.\nTraditionally, the localization problem has been tackled using 3D geometry.\nRecently, end-to-end approaches based on convolutional neural networks have\nbecome popular. These methods learn to directly regress the camera pose from an\ninput image. However, they do not achieve the same level of pose accuracy as 3D\nstructure-based methods. To understand this behavior, we develop a theoretical\nmodel for camera pose regression. We use our model to predict failure cases for\npose regression techniques and verify our predictions through experiments. We\nfurthermore use our model to show that pose regression is more closely related\nto pose approximation via image retrieval than to accurate pose estimation via\n3D structure. A key result is that current approaches do not consistently\noutperform a handcrafted image retrieval baseline. This clearly shows that\nadditional research is needed before pose regression algorithms are ready to\ncompete with structure-based methods.","url_abs":"http://arxiv.org/abs/1903.07504v1","url_pdf":"http://arxiv.org/pdf/1903.07504v1.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":"understanding-the-limitations-of-cnn-based","repo_url":"https://github.com/tsattler/understanding_apr","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"3d-geometry","task_name":"3D geometry"},{"task_slug":"camera-pose-estimation","task_name":"Camera Pose Estimation"},{"task_slug":"image-retrieval","task_name":"Image Retrieval"},{"task_slug":"mixed-reality","task_name":"Mixed Reality"},{"task_slug":"pose-estimation","task_name":"Pose Estimation"},{"task_slug":"retrieval","task_name":"Retrieval"},{"task_slug":"self-driving-cars","task_name":"Self-Driving Cars"},{"task_slug":"visual-localization","task_name":"Visual Localization"},{"task_slug":"regression-1","task_name":"regression"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1903.07504","atlas_url":"https://app.syntology.ai/?focus=1903.07504","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"1903.07504"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-25T09:33:49+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/tsattler/understanding_apr","reach":null}],"summary":{"ran_violates":1},"by_repo_kind":{"official":{"samples":1,"ran":1,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":1,"samples":[{"code_sha256_prefix":"f8ad33b3f20b644a","entry":"pair_id_to_image_ids","repo":"tsattler/understanding_apr","repo_kind":"official","path":"utils/export_image_names_and_ids.py","file_url":"https://github.com/tsattler/understanding_apr/blob/HEAD/utils/export_image_names_and_ids.py","link_basis":"first_harvest_node","language":"python","status":"ran_violates","verification_level":1,"contract_check":"VIOLATES","metamorphic_tier":"well_formed","behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"f8ad33b3f20b644a"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}