{"url":"/method/dynamic-smoothl1-loss","slug":"dynamic-smoothl1-loss","name":"Dynamic SmoothL1 Loss","full_name":"Dynamic SmoothL1 Loss","full_name_withheld":false,"description_markdown":"**Dynamic SmoothL1 Loss (DSL)** is a loss function in object detection where we change the shape of loss function to gradually focus on high quality samples:\r\n\r\n$$\\text{DSL}\\left(x, \\beta\\_{now}\\right) = 0.5|{x}|^{2}/\\beta\\_{now}, \\text{ if } |x| < \\beta\\_{now}\\text{,} $$ \r\n$$\\text{DSL}\\left(x, \\beta\\_{now}\\right) = |{x}| - 0.5\\beta\\_{now}\\text{, otherwise} $$ \r\n\r\nDSL will change the value of $\\beta\\_{now}$ according to the statistics of regression errors which can reflect the localization accuracy. It was introduced as part of the [Dynamic R-CNN](https://paperswithcode.com/method/dynamic-r-cnn) model.","description_state":"present","introduced_year":null,"introduced_by":{"title":null,"paper":null,"first_author":null,"n_authors":0,"url_abs":null,"archive_paper_url":null},"source":{"url":"https://arxiv.org/abs/2004.06002v2","title":"Dynamic R-CNN: Towards High Quality Object Detection via Dynamic Training","url_on_a_paper_host":true},"code_snippet_url":"https://github.com/hkzhang95/DynamicRCNN/blob/5b6cbe552231c7dbf2cdbd139369c313f16d2a72/dynamic_rcnn/det_opr/loss.py#L10","code_snippet_url_on_a_code_host":true,"categories":[{"area":"General","area_id":"general","collection":"Loss Functions","url":"/methods/category/loss-functions","pwc_aliases":[]}],"n_papers_tagged":3,"archive_num_papers":null,"papers_newest_first":[{"paper":null,"title":"Enhancing Tree Type Detection in Forest Fire Risk Assessment: Multi-Stage Approach and Color Encoding with Forest Fire Risk Evaluation Framework for UAV Imagery","date":"2024-07-27","arxiv_id":"2407.19184","n_code_links":0,"syntology":null},{"paper":null,"title":"Fracture Detection in Wrist X-ray Images Using Deep Learning-Based Object Detection Models","date":"2021-11-14","arxiv_id":"2111.07355","n_code_links":0,"syntology":null},{"paper":"/paper/dynamic-r-cnn-towards-high-quality-object","title":"Dynamic R-CNN: Towards High Quality Object Detection via Dynamic Training","date":"2020-04-13","arxiv_id":"2004.06002","n_code_links":3,"syntology":{"ran":2,"of":18,"unverified":16,"pointer_only":0}}],"papers_shown":3,"tasks":[{"task":"/task/object-detection","name":"Object Detection","papers":3},{"task":"/task/object-detection-1","name":"object-detection","papers":3},{"task":"/task/ensemble-learning","name":"Ensemble Learning","papers":1},{"task":"/task/fire-detection","name":"Fire Detection","papers":1},{"task":"/task/fracture-detection","name":"Fracture detection","papers":1},{"task":"/task/management","name":"Management","papers":1},{"task":"/task/medical-object-detection","name":"Medical Object Detection","papers":1},{"task":"/task/transfer-learning","name":"Transfer Learning","papers":1},{"task":"/task/high","name":"Vocal Bursts Intensity Prediction","papers":1},{"task":"/task/regression-1","name":"regression","papers":1}],"tasks_shown":10,"n_tasks":10,"usage_by_year":[{"year":"2020","papers":1},{"year":"2021","papers":1},{"year":"2024","papers":1}],"row_source":"embedded","archive":{"source":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","archive_url":"https://paperswithcode.com/method/dynamic-smoothl1-loss"},"syntology_read_at":"2026-09-24T18:15:14+00:00"}