{"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/learning-detection-with-diverse-proposals","title":"Learning Detection with Diverse Proposals","arxiv_id":"1704.03533","date":"2017-04-11","proceeding":"CVPR 2017 7","authors":["Samaneh Azadi","Jiashi Feng","Trevor Darrell"],"abstract":"To predict a set of diverse and informative proposals with enriched\nrepresentations, this paper introduces a differentiable Determinantal Point\nProcess (DPP) layer that is able to augment the object detection architectures.\nMost modern object detection architectures, such as Faster R-CNN, learn to\nlocalize objects by minimizing deviations from the ground-truth but ignore\ncorrelation between multiple proposals and object categories. Non-Maximum\nSuppression (NMS) as a widely used proposal pruning scheme ignores label- and\ninstance-level relations between object candidates resulting in multi-labeled\ndetections. In the multi-class case, NMS selects boxes with the largest\nprediction scores ignoring the semantic relation between categories of\npotential election. In contrast, our trainable DPP layer, allowing for Learning\nDetection with Diverse Proposals (LDDP), considers both label-level contextual\ninformation and spatial layout relationships between proposals without\nincreasing the number of parameters of the network, and thus improves location\nand category specifications of final detected bounding boxes substantially\nduring both training and inference schemes. Furthermore, we show that LDDP\nkeeps it superiority over Faster R-CNN even if the number of proposals\ngenerated by LDPP is only ~30% as many as those for Faster R-CNN.","url_abs":"http://arxiv.org/abs/1704.03533v1","url_pdf":"http://arxiv.org/pdf/1704.03533v1.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":"learning-detection-with-diverse-proposals","repo_url":"https://github.com/azadis/LDDP","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":{"status":"unanswered"}}],"tasks":[{"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":"pruning","method_name":"Pruning"},{"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":"https://app.syntology.ai/?focus=1704.03533","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}