{"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/multi-scale-location-aware-kernel","title":"Multi-scale Location-aware Kernel Representation for Object Detection","arxiv_id":"1804.00428","date":"2018-04-02","proceeding":"CVPR 2018 6","authors":["Hao Wang","Qilong Wang","Mingqi Gao","Peihua Li","WangMeng Zuo"],"abstract":"Although Faster R-CNN and its variants have shown promising performance in\nobject detection, they only exploit simple first-order representation of object\nproposals for final classification and regression. Recent classification\nmethods demonstrate that the integration of high-order statistics into deep\nconvolutional neural networks can achieve impressive improvement, but their\ngoal is to model whole images by discarding location information so that they\ncannot be directly adopted to object detection. In this paper, we make an\nattempt to exploit high-order statistics in object detection, aiming at\ngenerating more discriminative representations for proposals to enhance the\nperformance of detectors. To this end, we propose a novel Multi-scale\nLocation-aware Kernel Representation (MLKP) to capture high-order statistics of\ndeep features in proposals. Our MLKP can be efficiently computed on a modified\nmulti-scale feature map using a low-dimensional polynomial kernel\napproximation.Moreover, different from existing orderless global\nrepresentations based on high-order statistics, our proposed MLKP is location\nretentive and sensitive so that it can be flexibly adopted to object detection.\nThrough integrating into Faster R-CNN schema, the proposed MLKP achieves very\ncompetitive performance with state-of-the-art methods, and improves Faster\nR-CNN by 4.9% (mAP), 4.7% (mAP) and 5.0% (AP at IOU=[0.5:0.05:0.95]) on PASCAL\nVOC 2007, VOC 2012 and MS COCO benchmarks, respectively. Code is available at:\nhttps://github.com/Hwang64/MLKP.","url_abs":"http://arxiv.org/abs/1804.00428v1","url_pdf":"http://arxiv.org/pdf/1804.00428v1.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":"multi-scale-location-aware-kernel","repo_url":"https://github.com/Hwang64/MLKP","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"caffe2","reach":null},{"paper_slug":"multi-scale-location-aware-kernel","repo_url":"https://github.com/qilei123/MLKP4OpenImage","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"caffe2","reach":null}],"tasks":[{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"object","task_name":"Object"},{"task_slug":"object-detection","task_name":"Object Detection"},{"task_slug":"object-detection-1","task_name":"object-detection"}],"methods":[{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"faster-r-cnn","method_name":"Faster R-CNN"},{"method_slug":"rpn","method_name":"RPN"},{"method_slug":"roipool","method_name":"RoIPool"},{"method_slug":"softmax","method_name":"Softmax"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}