# KD-DETR: Knowledge Distillation for Detection Transformer with Consistent Distillation Points Sampling

arXiv: https://arxiv.org/abs/2211.08071 · HTML: https://mlsanity.com/paper/2211.08071/

## Key facts

- Cited by 13 papers in the map (data through 14 Oct 2026).
- Builds on 34 of its 43 references that are in the map.
- First posted on arXiv 15 Nov 2022.

## Builds on (most cited first, 30 of 34)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [Microsoft COCO: Common Objects in Context](https://mlsanity.com/paper/1405.0312/) (2014)
- [Distilling the Knowledge in a Neural Network](https://mlsanity.com/paper/1503.02531/) (2015)
- [Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks](https://mlsanity.com/paper/1506.01497/) (2015)
- [End-to-End Object Detection with Transformers](https://mlsanity.com/paper/2005.12872/) (2020)
- [You Only Look Once: Unified, Real-Time Object Detection](https://mlsanity.com/paper/1506.02640/) (2015)
- [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://mlsanity.com/paper/2010.04159/) (2020)
- [FitNets: Hints for Thin Deep Nets](https://mlsanity.com/paper/1412.6550/) (2014)
- [FCOS: Fully Convolutional One-Stage Object Detection](https://mlsanity.com/paper/1904.01355/) (2019)
- [DINO: DETR with Improved DeNoising Anchor Boxes for End-to-End Object Detection](https://mlsanity.com/paper/2203.03605/) (2022)
- [Paying More Attention to Attention: Improving the Performance of Convolutional Neural Networks via Attention Transfer](https://mlsanity.com/paper/1612.03928/) (2016)
- [Contrastive Representation Distillation](https://mlsanity.com/paper/1910.10699/) (2019)
- [DAB-DETR: Dynamic Anchor Boxes are Better Queries for DETR](https://mlsanity.com/paper/2201.12329/) (2022)
- [DN-DETR: Accelerate DETR Training by Introducing Query DeNoising](https://mlsanity.com/paper/2203.01305/) (2022)
- [Cascade R-CNN: High Quality Object Detection and Instance Segmentation](https://mlsanity.com/paper/1906.09756/) (2019)
- [Conditional DETR for Fast Training Convergence](https://mlsanity.com/paper/2108.06152/) (2021)
- [Generalized Focal Loss: Learning Qualified and Distributed Bounding Boxes for Dense Object Detection](https://mlsanity.com/paper/2006.04388/) (2020)
- [Decoupled Knowledge Distillation](https://mlsanity.com/paper/2203.08679/) (2022)
- [A Comprehensive Overhaul of Feature Distillation](https://mlsanity.com/paper/1904.01866/) (2019)
- [Libra R-CNN: Towards Balanced Learning for Object Detection](https://mlsanity.com/paper/1904.02701/) (2019)
- [Distilling Object Detectors with Fine-grained Feature Imitation](https://mlsanity.com/paper/1906.03609/) (2019)
- [Meta R-CNN : Towards General Solver for Instance-level Few-shot Learning](https://mlsanity.com/paper/1909.13032/) (2019)
- [Fast Convergence of DETR with Spatially Modulated Co-Attention](https://mlsanity.com/paper/2101.07448/) (2021)
- [Rethinking Transformer-based Set Prediction for Object Detection](https://mlsanity.com/paper/2011.10881/) (2020)
- [Efficient DETR: Improving End-to-End Object Detector with Dense Prior](https://mlsanity.com/paper/2104.01318/) (2021)
- [Cross-Image Relational Knowledge Distillation for Semantic Segmentation](https://mlsanity.com/paper/2204.06986/) (2022)
- [General Instance Distillation for Object Detection](https://mlsanity.com/paper/2103.02340/) (2021)
- [Masked Generative Distillation](https://mlsanity.com/paper/2205.01529/) (2022)
- [Distilling Object Detectors via Decoupled Features](https://mlsanity.com/paper/2103.14475/) (2021)
- [Sparse DETR: Efficient End-to-End Object Detection with Learnable Sparsity](https://mlsanity.com/paper/2111.14330/) (2021)

## Built on by (most cited first, 13 of 13)

- [D-FINE: Redefine Regression Task in DETRs as Fine-grained Distribution Refinement](https://mlsanity.com/paper/2410.13842/) (2024)
- [CLoCKDistill: Consistent Location-and-Context-aware Knowledge Distillation for DETRs](https://mlsanity.com/paper/2502.10683/) (2025)
- [Enhancing Indoor Occupancy Prediction via Sparse Query-Based Multi-Level Consistent Knowledge Distillation](https://mlsanity.com/paper/2602.02318/) (2026)
- [SO-DETR: Leveraging Dual-Domain Features and Knowledge Distillation for Small Object Detection](https://mlsanity.com/paper/2504.11470/) (2025)
- [Mr. DETR++: Instructive Multi-Route Training for Detection Transformers with Mixture-of-Experts](https://mlsanity.com/paper/2412.10028/) (2024)
- [Architectural Insights into Knowledge Distillation for Object Detection: A Comprehensive Review](https://mlsanity.com/paper/2508.03317/) (2025)
- [Symbiosis-Inspired Knowledge Distillation for Incremental Object Detection](https://mlsanity.com/paper/2607.13452/) (2026)
- [Decompose, Mix, Adapt: A Unified Framework for Parameter-Efficient Neural Network Recombination and Compression](https://mlsanity.com/paper/2603.27383/) (2026)
- [AnyDepth-DETR/-YOLO: Any-depth object detection with a single network](https://mlsanity.com/paper/2605.09407/) (2026)
- [M^2C-EvDet: Multi-Domain Multi-Order Cross-Modal Knowledge Distillation for Event-based Object Detection](https://mlsanity.com/paper/2606.24248/) (2026)
- [iFAN: Inference-Aware Learning for Plain Mask Transformers](https://mlsanity.com/paper/2608.03216/) (2026)
- [Enhanced Knowledge Distillation for Detection Transformer via Teacher Prediction Refinement](https://mlsanity.com/paper/2609.19964/) (2026)
- [Task-Sensitive Geometry of Representation Transfer for Object Detection under Image Degradation](https://mlsanity.com/paper/2610.04627/) (2026)

## Abstract (arXiv)

DETR is a novel end-to-end transformer architecture object detector, which significantly outperforms classic detectors when scaling up. In this paper, we focus on the compression of DETR with knowledge distillation. While knowledge distillation has been well-studied in classic detectors, there is a lack of researches on how to make it work effectively on DETR. We first provide experimental and theoretical analysis to point out that the main challenge in DETR distillation is the lack of consistent distillation points. Distillation points refer to the corresponding inputs of the predictions for student to mimic, which have different formulations in CNN detector and DETR, and reliable distillation requires sufficient distillation points which are consistent between teacher and student. Based on this observation, we propose the first general knowledge distillation paradigm for DETR (KD-DETR) with consistent distillation points sampling, for both homogeneous and heterogeneous distillation. Specifically, we decouple detection and distillation tasks by introducing a set of specialized object queries to construct distillation points for DETR. We further propose a general-to-specific distillation points sampling strategy to explore the extensibility of KD-DETR. Extensive experiments validate the effectiveness and generalization of KD-DETR. For both single-scale DAB-DETR and multis-scale Deformable DETR and DINO, KD-DETR boost the performance of student model with improvements of 2.6%-5.2%. We further extend KD-DETR to heterogeneous distillation, and achieves 2.1% improvement by distilling the knowledge from DINO to Faster R-CNN with ResNet-50, which is comparable with homogeneous distillation methods.The code is available at https://github.com/wennyuhey/KD-DETR.