mlsanityData to 14 Oct 2026

Unbiased Regression Loss for DETRs

arXiv 2410.22638 ↗First posted 30 Oct 2024cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

In this paper, we introduce a novel unbiased regression loss for DETR-based detectors. The conventional L1 regression loss tends to bias towards larger boxes, as they disproportionately contribute more towards the overall loss compared to smaller boxes. Consequently, the detection performance for small objects suffers. To alleviate this bias, the proposed new unbiased loss, termed Sized L1 loss, normalizes the size of all boxes based on their individual width and height. Our experiments demonstrate consistent improvements in both fully-supervised and semi-supervised settings using the MS-COCO benchmark dataset.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 25 of its 27 references that are in the map. First posted on arXiv 30 Oct 2024.

Builds on

25 of 27 references.

References · 25

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2014Microsoft COCO: Common Objects in Contextcited by 17,608
  3. 2015Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networkscited by 9,639
  4. 2020End-to-End Object Detection with Transformerscited by 7,414
  5. 2017Mask R-CNNcited by 7,277
  6. 2020Deformable DETR: Deformable Transformers for End-to-End Object Detectioncited by 3,422
  7. 2015Fast R-CNNcited by 2,933
  8. 2013Rich feature hierarchies for accurate object detection and semantic segmentationcited by 2,162
  9. 2019Generalized Intersection over Union: A Metric and A Loss for Bounding Box Regressioncited by 1,368
  10. 2022DINO: DETR with Improved DeNoising Anchor Boxes for End-to-End Object Detectioncited by 1,319
  11. 2022DAB-DETR: Dynamic Anchor Boxes are Better Queries for DETRcited by 573
  12. 2022DN-DETR: Accelerate DETR Training by Introducing Query DeNoisingcited by 488
13 more
  1. 2019Distance-IoU Loss: Faster and Better Learning for Bounding Box Regressioncited by 459
  2. 2021Conditional DETR for Fast Training Convergencecited by 424
  3. 2017Soft-NMS -- Improving Object Detection With One Line of Codecited by 382
  4. 2021End-to-End Semi-Supervised Object Detection with Soft Teachercited by 316
  5. 2021Unbiased Teacher for Semi-Supervised Object Detectioncited by 306
  6. 2020A Simple Semi-Supervised Learning Framework for Object Detectioncited by 282
  7. 2022DETRs with Collaborative Hybrid Assignments Trainingcited by 274
  8. 2019IoU Loss for 2D/3D Object Detectioncited by 128
  9. 2021Humble Teachers Teach Better Students for Semi-Supervised Object Detectioncited by 116
  10. 2021Instant-Teaching: An End-to-End Semi-Supervised Object Detection Frameworkcited by 113
  11. 2022PseCo: Pseudo Labeling and Consistency Training for Semi-Supervised Object Detectioncited by 51
  12. 2022Dense Teacher: Dense Pseudo-Labels for Semi-supervised Object Detectioncited by 48
  13. 2023Semi-DETR: Semi-Supervised Object Detection with Detection Transformerscited by 27

Built on by

Not cited yet.