{"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/a-unified-approach-of-multi-scale-deep-and","title":"A Unified Approach of Multi-scale Deep and Hand-crafted Features for Defocus Estimation","arxiv_id":"1704.08992","date":"2017-04-28","proceeding":"CVPR 2017 7","authors":["Jinsun Park","Yu-Wing Tai","Donghyeon Cho","In So Kweon"],"abstract":"In this paper, we introduce robust and synergetic hand-crafted features and a\nsimple but efficient deep feature from a convolutional neural network (CNN)\narchitecture for defocus estimation. This paper systematically analyzes the\neffectiveness of different features, and shows how each feature can compensate\nfor the weaknesses of other features when they are concatenated. For a full\ndefocus map estimation, we extract image patches on strong edges sparsely,\nafter which we use them for deep and hand-crafted feature extraction. In order\nto reduce the degree of patch-scale dependency, we also propose a multi-scale\npatch extraction strategy. A sparse defocus map is generated using a neural\nnetwork classifier followed by a probability-joint bilateral filter. The final\ndefocus map is obtained from the sparse defocus map with guidance from an\nedge-preserving filtered input image. Experimental results show that our\nalgorithm is superior to state-of-the-art algorithms in terms of defocus\nestimation. Our work can be used for applications such as segmentation, blur\nmagnification, all-in-focus image generation, and 3-D estimation.","url_abs":"http://arxiv.org/abs/1704.08992v1","url_pdf":"http://arxiv.org/pdf/1704.08992v1.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":"a-unified-approach-of-multi-scale-deep-and","repo_url":"https://github.com/zzangjinsun/DHDE_CVPR17","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"defocus-estimation","task_name":"Defocus Estimation"},{"task_slug":"image-generation","task_name":"Image Generation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/defocus-estimation-on-cuhk-blur-detection","task":"Defocus Estimation","dataset":"CUHK - Blur Detection Dataset","model":"DHDE","rank_in_archive_order":2,"of":5,"metrics":{"Blur Segmentation Accuracy":"83.73"},"uses_additional_data":false}],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=1704.08992","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}