mlsanityData to 7 Oct 2026

Focal Loss for Dense Object Detection

arXiv 1708.02002 ↗First posted 7 Aug 2017cs.CVcited by 749

Not on the map. Why?

Abstract (arXiv)

The highest accuracy object detectors to date are based on a two-stage approach popularized by R-CNN, where a classifier is applied to a sparse set of candidate object locations. In contrast, one-stage detectors that are applied over a regular, dense sampling of possible object locations have the potential to be faster and simpler, but have trailed the accuracy of two-stage detectors thus far. In this paper, we investigate why this is the case. We discover that the extreme foreground-background class imbalance encountered during training of dense detectors is the central cause. We propose to address this class imbalance by reshaping the standard cross entropy loss such that it down-weights the loss assigned to well-classified examples. Our novel Focal Loss focuses training on a sparse set of hard examples and prevents the vast number of easy negatives from overwhelming the detector during training. To evaluate the effectiveness of our loss, we design and train a simple dense detector we call RetinaNet. Our results show that when trained with the focal loss, RetinaNet is able to match the speed of previous one-stage detectors while surpassing the accuracy of all existing state-of-the-art two-stage detectors. Code is at: https://github.com/facebookresearch/Detectron.

Key facts. Cited by 749 papers in the map (data through 7 Oct 2026). First posted on arXiv 7 Aug 2017.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 261; 2022: 236; 2023: 221; 2024: 11; 2025: 11; 2026: 9.

749 papers.

Sort:

By area:

  1. 2023Segment Anythingcited by 8,158
  2. 2023Grounding DINO: Marrying DINO with Grounded Pre-Training for Open-Set Object Detectioncited by 3,274
  3. 2021Masked-attention Mask Transformer for Universal Image Segmentationcited by 2,338
  4. 2021Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutionscited by 1,960
  5. 2022DINO: DETR with Improved DeNoising Anchor Boxes for End-to-End Object Detectioncited by 1,306
  6. 2021Per-Pixel Classification is Not All You Need for Semantic Segmentationcited by 1,148
  7. 2021Grounded Language-Image Pre-trainingcited by 1,147
  8. 2022Planning-oriented Autonomous Drivingcited by 1,068
  9. 2021YOLOX: Exceeding YOLO Series in 2021cited by 1,064
  10. 2022BEVFusion: Multi-Task Multi-Sensor Fusion with Unified Bird's-Eye View Representationcited by 844
  11. 2021MDETR -- Modulated Detection for End-to-End Multi-Modal Understandingcited by 811
  12. 2021PVT v2: Improved Baselines with Pyramid Vision Transformercited by 749
38 more
  1. 2021RegionCLIP: Region-based Language-Image Pretrainingcited by 612
  2. 2022DAB-DETR: Dynamic Anchor Boxes are Better Queries for DETRcited by 569
  3. 2021Twins: Revisiting the Design of Spatial Attention in Vision Transformerscited by 565
  4. 2021DRAEM -- A discriminatively trained reconstruction embedding for surface anomaly detectioncited by 510
  5. 2022DN-DETR: Accelerate DETR Training by Introducing Query DeNoisingcited by 486
  6. 2022TransFusion: Robust LiDAR-Camera Fusion for 3D Object Detection with Transformerscited by 464
  7. 2022GraphMAE: Self-Supervised Masked Graph Autoencoderscited by 443
  8. 2021Tip-Adapter: Training-free CLIP-Adapter for Better Vision-Language Modelingcited by 429
  9. 2021Conditional DETR for Fast Training Convergencecited by 422
  10. 2021MOTR: End-to-End Multiple-Object Tracking with Transformercited by 379
  11. 2022PETRv2: A Unified Framework for 3D Perception from Multi-Camera Imagescited by 351
  12. 2023Faster Segment Anything: Towards Lightweight SAM for Mobile Applicationscited by 345
  13. 2022Mask DINO: Towards A Unified Transformer-based Framework for Object Detection and Segmentationcited by 341
  14. 2021Categorical Depth Distribution Network for Monocular 3D Object Detectioncited by 326
  15. 2021End-to-End Semi-Supervised Object Detection with Soft Teachercited by 315
  16. 2021Towards Open World Object Detectioncited by 297
  17. 2022Vision Transformer with Deformable Attentioncited by 293
  18. 2021Focal Self-attention for Local-Global Interactions in Vision Transformerscited by 286
  19. 2021Distilling Knowledge via Knowledge Reviewcited by 285
  20. 2022YOLOv6: A Single-Stage Object Detection Framework for Industrial Applicationscited by 273
  21. 2022DETRs with Collaborative Hybrid Assignments Trainingcited by 273
  22. 2021IconQA: A New Benchmark for Abstract Diagram Understanding and Visual Language Reasoningcited by 262
  23. 2022Transformers in Medical Imaging: A Surveycited by 245
  24. 2021Mobile-Former: Bridging MobileNet and Transformercited by 234
  25. 2023VoxelNeXt: Fully Sparse VoxelNet for 3D Object Detection and Trackingcited by 221
  26. 2021TOOD: Task-aligned One-stage Object Detectioncited by 218
  27. 2021Group-Free 3D Object Detection via Transformerscited by 215
  28. 2021K-Net: Towards Unified Image Segmentationcited by 211
  29. 2023Scene as Occupancycited by 208
  30. 2021Multi-Scale Vision Longformer: A New Vision Transformer for High-Resolution Image Encodingcited by 207
  31. 2021DetCo: Unsupervised Contrastive Learning for Object Detectioncited by 205
  32. 2023Universal Instance Perception as Object Discovery and Retrievalcited by 190
  33. 2021Dynamic Head: Unifying Object Detection Heads with Attentionscited by 185
  34. 2022Language as Queries for Referring Video Object Segmentationcited by 180
  35. 2023AnomalyGPT: Detecting Industrial Anomalies Using Large Vision-Language Modelscited by 169
  36. 2021Multimodal Virtual Point 3D Detectioncited by 166
  37. 2021Rethinking Rotated Object Detection with Gaussian Wasserstein Distance Losscited by 156
  38. 2022OnePose: One-Shot Object Pose Estimation without CAD Modelscited by 155

50 most cited of 749 All 749 (JSON)