Full search page →

RetinaNet and the focal loss

Optional · paper guide 5 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

Supplementary notes · 2

  1. Ultralytics releases without a paper
  2. Folding a re-parameterized block into one convolution

Paper guides · 22

  1. YOLO v1: detection as a single regression
  2. YOLOv2 and YOLO9000: anchors, priors and joint training
  3. YOLOv3: three scales and Darknet-53
  4. SSD: multi-scale default boxes
  5. RetinaNet and the focal loss
  6. FCOS: per-pixel anchor-free detection
  7. CenterNet: objects as points
  8. YOLOR: explicit features and learned implicit representations
  9. YOLOX: anchor-free YOLO with SimOTA
  10. YOLOv6: a re-parameterized industrial detector
  11. YOLOv7: a trainable bag of freebies
  12. YOLOv9: programmable gradient information
  13. YOLOv10: NMS-free training with dual assignments
  14. YOLOv12: an attention-centric YOLO
  15. YOLOv13: hypergraph correlation
  16. YOLO26: a DFL-free end-to-end YOLO
  17. RT-DETR: a real-time DETR
  18. RT-DETRv2: a bag of freebies for RT-DETR
  19. D-FINE: boxes as refined distributions
  20. DEIM: dense one-to-one matching
  21. RF-DETR: architecture search on a DINOv2 backbone
  22. DEIMv2: DINOv3 features for real-time DETRs

Deep Learning for Computer Vision

  1. 02 · Deep Image Classification
View the whole course tree →

Paper: T.-Y. Lin, P. Goyal, R. Girshick, K. He and P. Dollár, “Focal loss for dense object detection,” ICCV 2017 [11]. Related part of YOLOv4: box loss and the classification terms of the head; DLCV 03 defines the focal loss.

Challenge

In 2017, one-stage detectors such as SSD and YOLO were faster but less accurate than two-stage detectors. The paper argues that the main reason is class imbalance during training: a dense detector evaluates about 100,000 candidate locations per image, almost all easy background, and their many small losses dominate training [11]. Two-stage detectors avoid this with proposals and sampled mini-batches (for example 1:3 positives to negatives); one-stage detectors used hard example mining [72], which discards most of the data.

Method

The feature pyramid supplies multiple scales to shared classification and regression subnetworks; focal loss changes classification training rather than this pyramid's geometry.
Original paper figure 3 — The feature pyramid supplies multiple scales to shared classification and regression subnetworks; focal loss changes classification training rather than this pyramid's geometry.

Figure source: [11], Published paper PDF, PDF page 5. Original PDF.

Focal loss. For a binary classification with predicted probability pp for the true class and pt=pp_t = p if the label is positive and 1−p1 - p otherwise,

FL(pt)=−αt (1−pt)γlog⁡(pt),\text{FL}(p_t) = -\alpha_t\,(1 - p_t)^{\gamma} \log(p_t),

where γ≥0\gamma \ge 0 is the focusing parameter and αt\alpha_t a class-balancing weight (α\alpha for positives, 1−α1 - \alpha for negatives) [11]. With γ=0\gamma = 0 this is the balanced cross-entropy. The factor (1−pt)γ(1 - p_t)^\gamma shrinks the loss of well-classified examples: with γ=2\gamma = 2, an example with pt=0.9p_t = 0.9 gets 100 times less loss than with plain cross-entropy, so training focuses on hard examples. The paper uses γ=2\gamma = 2 and α=0.25\alpha = 0.25 [11].

RetinaNet. A ResNet [32] backbone with an FPN [33] on levels P3P_3 to P7P_7; at each level, anchors of areas 32232^2 to 5122512^2 with three aspect ratios {1:2,1:1,2:1}\{1{:}2, 1{:}1, 2{:}1\} and three scales {20,21/3,22/3}\{2^0, 2^{1/3}, 2^{2/3}\}, so A=9A = 9 anchors per location [11]. Anchors with IoU at least 0.5 with a ground-truth box are foreground, those below 0.4 are background, and the rest are ignored. Two small fully convolutional subnets run on every level with shared weights across levels: a classification subnet (four 3×33 \times 3 convolutions with 256 channels and ReLU, then a 3×33 \times 3 convolution with KAKA sigmoid outputs) and a class-agnostic box subnet with the same structure and 4A4A outputs [11]. The focal loss is summed over all anchors and normalized by the number of positive anchors.

Initialization. If all anchors start with a probability around 0.5, the many background anchors produce a huge loss at the first iterations. The bias of the final classification layer is therefore initialized to b=−log⁡((1−π)/π)b = -\log((1 - \pi)/\pi) with prior π=0.01\pi = 0.01, so every anchor starts with a foreground probability of about 0.01 [11].

Inference. Per level, at most 1,000 top-scoring predictions above a confidence of 0.05 are decoded; the predictions of all levels are merged and NMS with threshold 0.5 is applied [11].

Separate an easy negative from a hard negative. For a background location that the model confidently classifies as background, the modulating factor makes its classification contribution small. A misclassified background location still receives a substantial gradient. The mechanism changes the relative weighting of examples; it neither removes the box-regression objective nor guarantees that the largest remaining losses correspond to useful annotations.

This distinction matters with noisy labels. A mislabeled object can remain hard indefinitely, so focusing training on hard examples may also emphasize annotation errors. The focal-loss parameters must therefore be interpreted together with the foreground/background balance and the annotation quality. At inference the loss is absent: the network still produces dense boxes and class scores, and NMS remains part of the detector. A training-only loss change can improve the deployed prediction without becoming an extra inference module.

Results

The best model, ResNet-101-FPN trained with scale jitter and 1.5 times longer, reaches 39.1% AP on COCO test-dev, ahead of all one-stage and two-stage detectors the paper compares with [11]. Table 1(e) gives the speed–accuracy trade-off on test-dev, timed on an Nvidia M40: ResNet-50 at 500 pixels reaches 32.5% AP in 72 ms, ResNet-101 at 800 pixels 37.8% AP in 198 ms [11].

Ablations

Table 1 of the paper (COCO minival, ResNet-50 at 600 pixels unless noted) [11]:

  • (a) Balanced cross-entropy: the best α\alpha gives 31.1% AP (α=0.75\alpha = 0.75).
  • (b) Focal loss: γ=2\gamma = 2, α=0.25\alpha = 0.25 gives 34.0% AP, a gain of 2.9 points with the same network; results are fairly robust to the exact values (γ=1\gamma = 1: 33.7%, γ=5\gamma = 5: 32.2%).
  • (c) Anchors: one scale and one ratio gives 30.3% AP; three scales and three ratios 34.0%; more saturates.
  • (d) Focal loss against hard example mining (ResNet-101): the best online hard example mining variant reaches 32.8% AP, the focal loss 36.0%.

What it proves: the loss, not the architecture, closes most of the gap to two-stage detectors.

Conclusion

RetinaNet showed that a dense one-stage detector can be the most accurate detector once the training loss handles the foreground–background imbalance, and it fixed the modern one-stage layout: an FPN neck and separate, shared classification and box subnets.

Problems

It is accurate but not real-time (198 ms for the most accurate model), its anchors and IoU thresholds are hand-designed, and NMS is still required. YOLOv3 reported that the focal loss did not help its design (see YOLOv3), and YOLOv4 mentions the focal loss only as a way to address imbalance in its survey; quality-aware variants of the focal loss, such as the varifocal loss [73], returned in YOLOv6 and DEIM. FCOS removed the anchors.

References

  1. 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
  2. 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
  3. G. Jocher, “Ultralytics YOLOv5,” software (first released 2020; v7.0 released 22 November 2022). release · documentation
  4. G. Jocher, A. Chaurasia and J. Qiu, “Ultralytics YOLOv8,” software, version 8.0.0, 2023 (no paper). docs · code
  5. G. Jocher and J. Qiu, “Ultralytics YOLO11,” software, version 11.0.0, 2024 (no paper). docs · code
  6. 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
  7. 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
  8. J. Redmon and A. Farhadi, “YOLO9000: Better, faster, stronger,” in Proc. IEEE CVPR, pp. 6517–6525, 2017. doi · arXiv
  9. J. Redmon and A. Farhadi, “YOLOv3: An incremental improvement,” tech. report, arXiv:1804.02767, 2018. arXiv
  10. 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
  11. 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
  12. 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
  13. X. Zhou, D. Wang, P. Krähenbühl, “Objects as Points,” arXiv preprint, 2019. source
  14. C. Wang, I. Yeh, H. M. Liao, “You Only Learn One Representation: Unified Network for Multiple Tasks,” arXiv preprint, 2021. source
  15. Z. Ge, S. Liu, F. Wang, Z. Li and J. Sun, “YOLOX: Exceeding YOLO series in 2021,” arXiv:2107.08430, 2021. arXiv
  16. 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
  17. 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
  18. 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
  19. 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
  20. 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
  21. 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
  22. 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
  23. 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
  24. 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
  25. 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
  26. S. Huang, Y. Hou, L. Liu, X. Yu and X. Shen, “Real-time object detection meets DINOv3,” arXiv:2509.20787, 2025 (revised 2026). arXiv
  27. A. Bochkovskiy, C.-Y. Wang and H.-Y. M. Liao, “YOLOv4: Optimal speed and accuracy of object detection,” arXiv:2004.10934, 2020. arXiv
  28. 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
  29. NVIDIA, “TensorRT documentation,” official documentation. link
  30. 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
  31. K. Simonyan, A. Zisserman, “Very Deep Convolutional Networks for Large-Scale Image Recognition,” arXiv preprint, 2014. source
  32. K. He, X. Zhang, S. Ren, J. Sun, “Deep Residual Learning for Image Recognition,” arXiv preprint, 2015. source
  33. 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
  34. 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
  35. 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
  36. 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
  37. A. Bochkovskiy and contributors, “Darknet: YOLOv4 configuration and detection layer,” official implementation, accessed 2 November 2026. configuration · assignment and decoding
  38. M. Tan, Q. V. Le, “EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks,” ICML, 2019. source · proceedings
  39. 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
  40. 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
  41. D. Misra, “Mish: A Self Regularized Non-Monotonic Activation Function,” BMVC, 2020. source
  42. P. Ramachandran, B. Zoph, Q. V. Le, “Searching for Activation Functions,” arXiv preprint, 2017. source
  43. J. Yu, Y. Jiang, Z. Wang, Z. Cao, T. Huang, “UnitBox: An Advanced Object Detection Network,” ACM Multimedia, 2016. source
  44. 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
  45. 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
  46. H. Zhang, M. Cisse, Y. N. Dauphin and D. Lopez-Paz, “mixup: Beyond empirical risk minimization,” in Proc. ICLR, 2018. arXiv
  47. 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
  48. 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
  49. G. Ghiasi, T. Lin, Q. V. Le, “DropBlock: A regularization method for convolutional networks,” NeurIPS, 2018. source · proceedings
  50. 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
  51. 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
  52. 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
  53. 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
  54. I. Loshchilov, F. Hutter, “SGDR: Stochastic Gradient Descent with Warm Restarts,” ICLR, 2017. source
  55. 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
  56. COCO Consortium, “COCO detection evaluation,” official evaluation description. link
  57. ONNX, “Open Neural Network Exchange,” official project page. link
  58. 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
  59. 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
  60. H. W. Kuhn, “The Hungarian method for the assignment problem,” Naval Research Logistics Quarterly, vol. 2, pp. 83–97, 1955. doi
  61. 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
  62. 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
  63. 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
  64. 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
  65. Ultralytics, “YOLO27 (coming soon),” official documentation page, accessed 2 November 2026. docs
  66. M. Oquab, T. Darcet, T. Moutakanni, H. Vo, M. Szafraniec et al., “DINOv2: Learning robust visual features without supervision,” arXiv:2304.07193, 2023. arXiv
  67. O. Siméoni, H. V. Vo, M. Seitzer, F. Baldassarre, M. Oquab et al., “DINOv3,” arXiv:2508.10104, 2025. arXiv
  68. 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
  69. R. Girshick, “Fast R-CNN,” 2015 IEEE International Conference on Computer Vision (ICCV), 2015. doi · manuscript
  70. S. Ren, K. He, R. Girshick, J. Sun, “Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks,” arXiv preprint, 2015. source
  71. 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
  72. 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
  73. 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
  74. YOLOX contributors, “YOLOX detection head,” official implementation, accessed 2 November 2026. code
  75. K. Duan, S. Bai, L. Xie, H. Qi, Q. Huang, Q. Tian, “CenterNet: Keypoint Triplets for Object Detection,” arXiv preprint, 2019. source
  76. 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
  77. C. Li, L. Li, Y. Geng, et al., “YOLOv6 v3.0: A Full-Scale Reloading,” arXiv preprint, 2023. source
  78. 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
  79. Z. Gevorgyan, “SIoU Loss: More Powerful Learning for Bounding Box Regression,” arXiv preprint, 2022. source
  80. 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
  81. X. Ding, H. Chen, X. Zhang, K. Huang, J. Han, G. Ding, “Re-parameterizing Your Optimizers rather than Architectures,” ICLR, 2023. source
  82. 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
  83. C. Wang, H. M. Liao, I. Yeh, “Designing Network Design Strategies Through Gradient Path Analysis,” arXiv preprint, 2022. source
  84. 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
  85. D. P. Kingma, J. Ba, “Adam: A Method for Stochastic Optimization,” ICLR, 2015. source
  86. K. Jordan, “Muon: An optimizer for hidden layers in neural networks,” technical blog, 2024. source
  87. 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