DEIMv2: DINOv3 features for real-time DETRs
Optional · paper guide 22 of 22 for DLCV 03A · One-Stage Detection, Read Through YOLOv4. Citation numbers refer to the reference list at the bottom of this page.
Navigate: this page and this track: Navigate: this page and this track
On this page
- YOLO v1: detection as a single regression
- YOLOv2 and YOLO9000: anchors, priors and joint training
- YOLOv3: three scales and Darknet-53
- SSD: multi-scale default boxes
- RetinaNet and the focal loss
- FCOS: per-pixel anchor-free detection
- CenterNet: objects as points
- YOLOR: explicit features and learned implicit representations
- YOLOX: anchor-free YOLO with SimOTA
- YOLOv6: a re-parameterized industrial detector
- YOLOv7: a trainable bag of freebies
- YOLOv9: programmable gradient information
- YOLOv10: NMS-free training with dual assignments
- YOLOv12: an attention-centric YOLO
- YOLOv13: hypergraph correlation
- YOLO26: a DFL-free end-to-end YOLO
- RT-DETR: a real-time DETR
- RT-DETRv2: a bag of freebies for RT-DETR
- D-FINE: boxes as refined distributions
- DEIM: dense one-to-one matching
- RF-DETR: architecture search on a DINOv2 backbone
- DEIMv2: DINOv3 features for real-time DETRs
Deep Learning for Computer Vision
Huang et al.’s Real-Time Object Detection Meets DINOv3 technical report develops a family of detectors spanning larger Transformer backbones and very small convolutional models [26]. The discussion here follows its revised 2026 manuscript; the first preprint appeared in 2025.
Challenge
DINOv3 [67] supplies strong pretrained visual features, but a plain vision Transformer naturally produces a single main spatial resolution. Detection needs both semantics and fine detail across object scales. Simply inserting a large backbone does not establish a useful real-time detector, especially for very small deployment budgets.
DEIMv2 therefore has two related challenges: adapt pretrained Transformer features efficiently for the larger models, and design a different, compact path for the smallest models. It does not force every size to use the same backbone family.
Method

Figure source: [26], arXiv:2509.20787v5, PDF page 3. Original PDF.
Larger variants. The L and X configurations use DINOv3 ViT-Small and ViT-Small+ backbones. The S and M configurations use smaller backbones distilled from DINOv3. They retain pretrained semantic structure while changing the hidden width and downstream detector to fit a smaller budget. Pretraining and distillation remain part of the training history, even though only the resulting student runs at inference.
Spatial Tuning Adapter. STA combines two information paths. Intermediate Transformer block features at stride 16 are resized to the scales required by detection. A lightweight convolutional path extracts local detail from the image through fast downsampling. The Bi-Fusion operation combines these detail features with resized semantic features using channel projections and lightweight operations.
For example, a feature tensor can be resized to , and before fusion. are the input dimensions and the channel count. Resizing itself can be parameter-free, but the complete adapter is not parameter-free: its detail path and projections contain learned parameters. Upsampling also does not recreate image details absent from the semantic features; the parallel path supplies complementary evidence.
Smallest variants. Nano, Pico, Femto and Atto instead use progressively pruned HGNetv2 convolutional backbones. The report removes stages, reduces block counts and reduces channels as the budget shrinks. Their queries, hidden dimensions and resolutions also differ. Calling every variant a DINOv3-powered detector would therefore misdescribe the actual family.
Decoder and training. The report simplifies decoder operations, including its feed-forward and normalization layers, and shares a positional embedding across decoder layers. The motivation is that iterative box refinement often changes query positions only modestly. Training builds on Dense O2O and MAL from DEIM and introduces Copy-Blend to combine inserted objects with existing content rather than fully overwrite regions.
Its full training objective combines classification, localization and distillation terms:
The are loss weights; MAL is Matchability-Aware Loss, FGL fine-grained localization supervision, DDF distributional distillation, a box-coordinate loss and GIoU the generalized overlap loss [24], [23], [44]. The revised report excludes FGL and DDF for Pico, Femto and Atto because these terms hurt its tested ultra-small models. The equation is a family template, not evidence that all terms are active for every size.
Results
Table 1 of the revised manuscript reports 57.8 AP / 50.26 million parameters / 13.75 ms for X, 56.0 / 32.18 / 10.47 for L, 53.0 / 18.11 / 8.80 for M and 50.9 / 9.71 / 5.78 for S [26]. Preserve the manuscript’s model and evaluation settings when using these values. The introduction contains an inconsistent 57.6 figure for X; the values here follow the explicit results table rather than silently mixing versions.
At the smallest end, Table 4 reports 38.5 AP for Pico at , 31.0 for Femto at and 23.8 for Atto at [26]. These different input sizes are part of the comparison. Parameter count alone does not capture the accuracy–latency trade-off.
Ablations
This technical report provides configuration and component observations rather than a complete factorial ablation of every change. It reports that removing FGL and DDF helps its ultra-small variants, and that Copy-Blend improves its tested setup. Without a fully controlled table for each ingredient, those statements should remain scoped to the reported experiments.
The scale-specific breakdown also tempers the headline: the report says gains from DINOv3-based models are strongest on medium and large objects, while small-object gains are more limited. A semantic backbone improvement should not automatically be described as recovering fine detail or solving tiny-object detection.
Conclusion
DEIMv2 shows how to adapt a strong visual representation while preserving a separate local-detail path, and when to use a different backbone for a much smaller budget. Its family design is more informative than treating a single largest-model AP as the whole contribution.
Problems
The report spans multiple manuscript revisions, and configuration names alone may hide changed results. Match the paper version, checkpoint and export settings. The authors also state that latency has not been fully optimized, so the listed timings should not be treated as a hardware-independent ranking.
Finally, distillation, backbone pretraining, adapter design and augmentation all influence the result. Reproducing only the adapter on an unrelated backbone does not reproduce DEIMv2. For a new application, test small-object errors, memory and the actual compiled graph before choosing a variant.
References
- C.-Y. Wang, A. Bochkovskiy and H.-Y. M. Liao, “YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors,” in Proc. IEEE/CVF CVPR, pp. 7464–7475, 2023. doi
- C.-Y. Wang, I-H. Yeh and H.-Y. M. Liao, “YOLOv9: Learning what you want to learn using programmable gradient information,” in Proc. ECCV, 2024. doi
- G. Jocher, “Ultralytics YOLOv5,” software (first released 2020; v7.0 released 22 November 2022). release · documentation
- G. Jocher, A. Chaurasia and J. Qiu, “Ultralytics YOLOv8,” software, version 8.0.0, 2023 (no paper). docs · code
- G. Jocher and J. Qiu, “Ultralytics YOLO11,” software, version 11.0.0, 2024 (no paper). docs · code
- J. Lu, J. Jia, J. Yawl and C. Zhang, “YOLOv14: Adaptive real-time object detection for diverse imaging conditions,” arXiv:2608.04720, 2026 (withdrawn by the authors, 20 August 2026). arXiv
- J. Redmon, S. Divvala, R. Girshick and A. Farhadi, “You only look once: Unified, real-time object detection,” in Proc. IEEE CVPR, pp. 779–788, 2016. doi · arXiv
- J. Redmon and A. Farhadi, “YOLO9000: Better, faster, stronger,” in Proc. IEEE CVPR, pp. 6517–6525, 2017. doi · arXiv
- J. Redmon and A. Farhadi, “YOLOv3: An incremental improvement,” tech. report, arXiv:1804.02767, 2018. arXiv
- W. Liu, D. Anguelov, D. Erhan, C. Szegedy, S. Reed, C. Fu, A. C. Berg, “SSD: Single Shot MultiBox Detector,” Lecture Notes in Computer Science / Computer Vision – ECCV 2016, 2016. doi · manuscript
- T.-Y. Lin, P. Goyal, R. Girshick, K. He and P. Dollár, “Focal loss for dense object detection,” in Proc. IEEE ICCV, pp. 2999–3007, 2017. doi
- Z. Tian, C. Shen, H. Chen, T. He, “FCOS: Fully Convolutional One-Stage Object Detection,” 2019 IEEE/CVF International Conference on Computer Vision (ICCV), 2019. doi · manuscript
- X. Zhou, D. Wang, P. Krähenbühl, “Objects as Points,” arXiv preprint, 2019. source
- C. Wang, I. Yeh, H. M. Liao, “You Only Learn One Representation: Unified Network for Multiple Tasks,” arXiv preprint, 2021. source
- Z. Ge, S. Liu, F. Wang, Z. Li and J. Sun, “YOLOX: Exceeding YOLO series in 2021,” arXiv:2107.08430, 2021. arXiv
- C. Li, L. Li, H. Jiang, K. Weng, Y. Geng et al., “YOLOv6: A single-stage object detection framework for industrial applications,” tech. report, arXiv:2209.02976, 2022. arXiv
- A. Wang, H. Chen, L. Liu, K. Chen, Z. Lin, J. Han and G. Ding, “YOLOv10: Real-time end-to-end object detection,” in Proc. NeurIPS, 2024. arXiv
- Y. Tian, Q. Ye and D. Doermann, “YOLOv12: Attention-centric real-time object detectors,” in Advances in Neural Information Processing Systems 38 (NeurIPS), 2025. proceedings · arXiv
- M. Lei, S. Li, Y. Wu, H. Hu, Y. Zhou et al., “YOLOv13: Real-time object detection with hypergraph-enhanced adaptive visual perception,” arXiv:2506.17733, 2025. arXiv
- G. Jocher, J. Qiu, M. Liu, S. Lyu, F. C. Akyon and M. E. Kalfaoglu, “Ultralytics YOLO26: Unified real-time end-to-end vision models,” arXiv:2606.03748, 2026 (software released January 2026). arXiv · docs
- Y. Zhao, W. Lv, S. Xu, J. Wei, G. Wang, Q. Dang, Y. Liu, J. Chen, “DETRs Beat YOLOs on Real-time Object Detection,” Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024. source
- W. Lv, Y. Zhao, Q. Chang, K. Huang, G. Wang and Y. Liu, “RT-DETRv2: Improved baseline with bag-of-freebies for real-time detection transformer,” arXiv:2407.17140, 2024. arXiv
- Y. Peng, H. Li, P. Wu, Y. Zhang, X. Sun and F. Wu, “D-FINE: Redefine regression task in DETRs as fine-grained distribution refinement,” in Proc. ICLR, 2025. arXiv · code
- S. Huang, Z. Lu, X. Cun, Y. Yu, X. Zhou, X. Shen, “DEIM: DETR with Improved Matching for Fast Convergence,” Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2025. source
- I. Robinson, P. Robicheaux, M. Popov, D. Ramanan and N. Peri, “RF-DETR: Neural architecture search for real-time detection transformers,” in Proc. ICLR, 2026. arXiv
- S. Huang, Y. Hou, L. Liu, X. Yu and X. Shen, “Real-time object detection meets DINOv3,” arXiv:2509.20787, 2025 (revised 2026). arXiv
- A. Bochkovskiy, C.-Y. Wang and H.-Y. M. Liao, “YOLOv4: Optimal speed and accuracy of object detection,” arXiv:2004.10934, 2020. arXiv
- M. Tan, R. Pang, Q. V. Le, “EfficientDet: Scalable and Efficient Object Detection,” 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020. doi · manuscript
- NVIDIA, “TensorRT documentation,” official documentation. link
- J. Deng, W. Dong, R. Socher, L. Li, Kai Li, Li Fei-Fei, “ImageNet: A large-scale hierarchical image database,” 2009 IEEE Conference on Computer Vision and Pattern Recognition, 2009. source
- K. Simonyan, A. Zisserman, “Very Deep Convolutional Networks for Large-Scale Image Recognition,” arXiv preprint, 2014. source
- K. He, X. Zhang, S. Ren, J. Sun, “Deep Residual Learning for Image Recognition,” arXiv preprint, 2015. source
- T.-Y. Lin, P. Dollár, R. Girshick, K. He, B. Hariharan and S. Belongie, “Feature pyramid networks for object detection,” in Proc. IEEE CVPR, 2017. doi
- S. Liu, L. Qi, H. Qin, J. Shi and J. Jia, “Path aggregation network for instance segmentation,” in Proc. IEEE/CVF CVPR, pp. 8759–8768, 2018. doi
- K. He, X. Zhang, S. Ren, J. Sun, “Spatial Pyramid Pooling in Deep Convolutional Networks for Visual Recognition,” Lecture Notes in Computer Science / Computer Vision – ECCV 2014, 2014. doi · manuscript
- C.-Y. Wang, H.-Y. M. Liao, I-H. Yeh, Y.-H. Wu, P.-Y. Chen and J.-W. Hsieh, “CSPNet: A new backbone that can enhance learning capability of CNN,” CVPR Workshops, 2020. arXiv · doi
- A. Bochkovskiy and contributors, “Darknet: YOLOv4 configuration and detection layer,” official implementation, accessed 2 November 2026. configuration · assignment and decoding
- M. Tan, Q. V. Le, “EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks,” ICML, 2019. source · proceedings
- G. Huang, Z. Liu, L. Van Der Maaten, K. Q. Weinberger, “Densely Connected Convolutional Networks,” 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017. doi · manuscript
- S. Xie, R. Girshick, P. Dollar, Z. Tu, K. He, “Aggregated Residual Transformations for Deep Neural Networks,” 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017. doi · manuscript
- D. Misra, “Mish: A Self Regularized Non-Monotonic Activation Function,” BMVC, 2020. source
- P. Ramachandran, B. Zoph, Q. V. Le, “Searching for Activation Functions,” arXiv preprint, 2017. source
- J. Yu, Y. Jiang, Z. Wang, Z. Cao, T. Huang, “UnitBox: An Advanced Object Detection Network,” ACM Multimedia, 2016. source
- H. Rezatofighi, N. Tsoi, J. Gwak, A. Sadeghian, I. Reid, S. Savarese, “Generalized Intersection over Union: A Metric and A Loss for Bounding Box Regression,” arXiv preprint, 2019. source
- Z. Zheng, P. Wang, W. Liu, J. Li, R. Ye and D. Ren, “Distance-IoU loss: Faster and better learning for bounding box regression,” in Proc. AAAI, pp. 12993–13000, 2020. doi
- H. Zhang, M. Cisse, Y. N. Dauphin and D. Lopez-Paz, “mixup: Beyond empirical risk minimization,” in Proc. ICLR, 2018. arXiv
- S. Yun, D. Han, S. Chun, S. J. Oh, Y. Yoo, J. Choe, “CutMix: Regularization Strategy to Train Strong Classifiers With Localizable Features,” 2019 IEEE/CVF International Conference on Computer Vision (ICCV), 2019. doi · manuscript
- Nitish Srivastava, Geoffrey Hinton, Alex Krizhevsky, Ilya Sutskever, Ruslan Salakhutdinov, “Dropout: A Simple Way to Prevent Neural Networks from Overfitting,” Journal of Machine Learning Research, 2014. source
- G. Ghiasi, T. Lin, Q. V. Le, “DropBlock: A regularization method for convolutional networks,” NeurIPS, 2018. source · proceedings
- C. Szegedy, V. Vanhoucke, S. Ioffe, J. Shlens, Z. Wojna, “Rethinking the Inception Architecture for Computer Vision,” 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016. doi · manuscript
- S. Ioffe and C. Szegedy, “Batch normalization: Accelerating deep network training by reducing internal covariate shift,” in Proc. ICML, PMLR vol. 37, pp. 448–456, 2015. link
- Z. Yao, Y. Cao, S. Zheng, G. Huang, S. Lin, “Cross-Iteration Batch Normalization,” 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021. doi · manuscript
- S. Woo, J. Park, J. Lee, I. S. Kweon, “CBAM: Convolutional Block Attention Module,” Lecture Notes in Computer Science / Computer Vision – ECCV 2018, 2018. doi · manuscript
- I. Loshchilov, F. Hutter, “SGDR: Stochastic Gradient Descent with Warm Restarts,” ICLR, 2017. source
- T.-Y. Lin, M. Maire, S. Belongie, J. Hays, P. Perona, D. Ramanan, P. Dollár and C. L. Zitnick, “Microsoft COCO: Common objects in context,” in Proc. ECCV, pp. 740–755, 2014. doi
- COCO Consortium, “COCO detection evaluation,” official evaluation description. link
- ONNX, “Open Neural Network Exchange,” official project page. link
- B. Jacob, S. Kligys, B. Chen, M. Zhu, M. Tang, A. Howard, H. Adam and D. Kalenichenko, “Quantization and training of neural networks for efficient integer-arithmetic-only inference,” in Proc. IEEE/CVF CVPR, pp. 2704–2713, 2018. doi
- N. Carion, F. Massa, G. Synnaeve, N. Usunier, A. Kirillov and S. Zagoruyko, “End-to-end object detection with transformers,” in Proc. ECCV, 2020. doi
- H. W. Kuhn, “The Hungarian method for the assignment problem,” Naval Research Logistics Quarterly, vol. 2, pp. 83–97, 1955. doi
- J. Terven, D.-M. Córdova-Esparza and J.-A. Romero-González, “A comprehensive review of YOLO architectures in computer vision: From YOLOv1 to YOLOv8 and YOLO-NAS,” Machine Learning and Knowledge Extraction, vol. 5, no. 4, pp. 1680–1716, 2023. doi · arXiv
- N. Jegham, C. Y. Koh, M. Abdelatti and A. Hendawi, “YOLO evolution: A comprehensive benchmark and architectural review of YOLOv12, YOLO11, and their previous versions,” arXiv:2411.00201, 2024 (revised 2025). arXiv
- R. Sapkota and M. Karkee, “Ultralytics YOLO evolution: An overview of YOLO27, YOLO26, YOLO11, YOLOv8, and YOLOv5 object detectors for computer vision and pattern recognition,” arXiv:2510.09653, 2025 (revised 2026). arXiv
- T. Shehzadi, K. A. Hashmi, M. Liwicki, D. Stricker and M. Z. Afzal, “Object detection with transformers: A review,” Sensors, vol. 25, no. 19, 2025. doi
- Ultralytics, “YOLO27 (coming soon),” official documentation page, accessed 2 November 2026. docs
- M. Oquab, T. Darcet, T. Moutakanni, H. Vo, M. Szafraniec et al., “DINOv2: Learning robust visual features without supervision,” arXiv:2304.07193, 2023. arXiv
- O. Siméoni, H. V. Vo, M. Seitzer, F. Baldassarre, M. Oquab et al., “DINOv3,” arXiv:2508.10104, 2025. arXiv
- X. Ding, X. Zhang, N. Ma, J. Han, G. Ding and J. Sun, “RepVGG: Making VGG-style ConvNets great again,” in Proc. IEEE/CVF CVPR, pp. 13728–13737, 2021. doi
- R. Girshick, “Fast R-CNN,” 2015 IEEE International Conference on Computer Vision (ICCV), 2015. doi · manuscript
- S. Ren, K. He, R. Girshick, J. Sun, “Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks,” arXiv preprint, 2015. source
- M. Everingham, L. Van Gool, C. K. I. Williams, J. Winn, A. Zisserman, “The Pascal Visual Object Classes (VOC) Challenge,” International Journal of Computer Vision, 2009. source
- A. Shrivastava, A. Gupta, R. Girshick, “Training Region-Based Object Detectors with Online Hard Example Mining,” 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016. doi · manuscript
- H. Zhang, Y. Wang, F. Dayoub and N. Sünderhauf, “VarifocalNet: An IoU-aware dense object detector,” in Proc. IEEE/CVF CVPR, pp. 8510–8519, 2021. doi
- YOLOX contributors, “YOLOX detection head,” official implementation, accessed 2 November 2026. code
- K. Duan, S. Bai, L. Xie, H. Qi, Q. Huang, Q. Tian, “CenterNet: Keypoint Triplets for Object Detection,” arXiv preprint, 2019. source
- Z. Ge, S. Liu, Z. Li, O. Yoshie and J. Sun, “OTA: Optimal transport assignment for object detection,” in Proc. IEEE/CVF CVPR, pp. 303–312, 2021. doi
- C. Li, L. Li, Y. Geng, et al., “YOLOv6 v3.0: A Full-Scale Reloading,” arXiv preprint, 2023. source
- C. Feng, Y. Zhong, Y. Gao, M. R. Scott, W. Huang, “TOOD: Task-Aligned One-Stage Object Detection,” Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021. source
- Z. Gevorgyan, “SIoU Loss: More Powerful Learning for Bounding Box Regression,” arXiv preprint, 2022. source
- X. Li, W. Wang, L. Wu, S. Chen, X. Hu, J. Li, J. Tang and J. Yang, “Generalized focal loss: Learning qualified and distributed bounding boxes for dense object detection,” in Advances in Neural Information Processing Systems 33 (NeurIPS), 2020. proceedings
- X. Ding, H. Chen, X. Zhang, K. Huang, J. Han, G. Ding, “Re-parameterizing Your Optimizers rather than Architectures,” ICLR, 2023. source
- S. Zhang, C. Chi, Y. Yao, Z. Lei, S. Z. Li, “Bridging the Gap Between Anchor-Based and Anchor-Free Detection via Adaptive Training Sample Selection,” 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020. doi · manuscript
- C. Wang, H. M. Liao, I. Yeh, “Designing Network Design Strategies Through Gradient Path Analysis,” arXiv preprint, 2022. source
- T. Dao, D. Fu, S. Ermon, A. Rudra, C. Ré, “FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness,” Advances in Neural Information Processing Systems 35, 2022. doi · manuscript
- D. P. Kingma, J. Ba, “Adam: A Method for Stochastic Optimization,” ICLR, 2015. source
- K. Jordan, “Muon: An optimizer for hidden layers in neural networks,” technical blog, 2024. source
- S. Shao, Z. Li, T. Zhang, C. Peng, G. Yu, X. Zhang, J. Li, J. Sun, “Objects365: A Large-Scale, High-Quality Dataset for Object Detection,” Proceedings of the IEEE/CVF International Conference on Computer Vision, 2019. source