The DOI matches its Crossref record; 47 of the 49 references it cites were found in a registry or an index.
Active Learning for Deep Object Detection via Probabilistic Modeling
Open accessAbstract
Active learning aims to reduce labeling costs by selecting only the most informative samples on a dataset. Few existing works have addressed active learning for object detection. Most of these methods are based on multiple models or are straightforward extensions of classification methods, hence estimate an image’s informativeness using only the classification head. In this paper, we propose a novel deep active learning approach for object detection. Our approach relies on mixture density networks that estimate a probabilistic distribution for each localization and classification head’s output. We explicitly estimate the aleatoric and epistemic uncertainty in a single forward pass of a single model. Our method uses a scoring function that aggregates these two types of uncertainties for both heads to obtain every image’s informativeness score. We demonstrate the efficacy of our approach in PASCAL VOC and MS-COCO datasets. Our approach outperforms single-model based methods and performs on par with multi-model based methods at a fraction of the computing cost. Code is available at https://github.com/NVlabs/AL-MDN.
Overview
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The cited source doesn’t support the claim as written.
In the citing paper
“In the last one, where we use 7 , 000 samples, 5 , 000 of which are actively labeled, our method outperforms the random baseline by 9 . 1% and the best existing active learning methods by more than 2 . 8% [5].”
From Not All Labels Are Equal: Rationalizing The Labeling Costs for Training Object Detection (2021) · cites this paper as [5]
Checked by a reasoning model — may be wrong.
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“For MS-COCO, in Fig. 4a-b, we observe that in the first active learning cycle, our method outperforms the random baseline by 5 . 8% , the best-performing AL method by 2 . 7% [5], the semi-supervised method by 5 . 4% , and the ensembles by 5% .”
From Not All Labels Are Equal: Rationalizing The Labeling Costs for Training Object Detection (2021) · cites this paper as [5]
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“We observe that this difference is maintained in the other cycles, including the last active learning cycle where our method outperforms the semisupervised method, multi-model methods [9, 12] and the best AL method [5] by 1 . 6% .”
From Not All Labels Are Equal: Rationalizing The Labeling Costs for Training Object Detection (2021) · cites this paper as [5]
The cited source says what the claim attributes to it.
In the citing paper
“In the second cycle, our approach outperforms all the other methods, including PM [5], by almost 4% or more.”
From Not All Labels Are Equal: Rationalizing The Labeling Costs for Training Object Detection (2021) · cites this paper as [5]
The cited source says what the claim attributes to it.
In the citing paper
“Compared to state-of-the-art methods [4-8], our approach yields up to a 7 . 7% and 7% relative mAP improvement for PASCAL-VOC and MS-COCO, respectively.”
From Not All Labels Are Equal: Rationalizing The Labeling Costs for Training Object Detection (2021) · cites this paper as [5]
The cited source says what the claim attributes to it.
In the citing paper
“The random sampling strategy is mostly used for comparison purposes, grounded on to the principle that AL-based methods have the potential to perform better (e.g., [5,48,52,60]).”
From Ten Years of Active Learning Techniques and Object Detection: A Systematic Review (2023) · cites this paper as [48]
The cited source says what the claim attributes to it.
In the citing paper
“The classifiers used for uncertainty estimation ignore the location information [48].”
From Ten Years of Active Learning Techniques and Object Detection: A Systematic Review (2023) · cites this paper as [48]
The cited source says what the claim attributes to it.
In the citing paper
“White methods, based on multi-models (like classifier assemblies which increase the number of parameters [48]), have a high computing cost, while black methods may not need many parameters.”
From Ten Years of Active Learning Techniques and Object Detection: A Systematic Review (2023) · cites this paper as [48]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474 · cited 1×
“We compare the performance of SSD using different backbones such as Resnet-34 and Resnet-50 [17], and Faster-RCNN [32] detector trained using our actively sampled dataset.”
MethodVery Deep Convolutional Networks for Large-Scale Image Recognition
Simonyan, K., & Zisserman, A. (2014) · arXiv · doi:10.48550/arxiv.1409.1556 · Cited by 114,148 · cited 1×
“Experimental settings: We employ Single Shot MultiBox Detector (SSD) [31], which is widely used in active learning studies [33, 40], with a VGG-16 backbone [37].”
MethodFaster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
Ren, S., He, K., Girshick, R., & Sun, J. (2017) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2016.2577031 · Cited by 73,472
A Mathematical Theory of Communication
Shannon, C. E. (1948) · The MIT Press eBooks · doi:10.1002/j.1538-7305.1948.tb00917.x · Cited by 50,387
Microsoft COCO: Common Objects in Context
Lin, T.-Y., Maire, M., Belongie, S., Hays, J., Perona, P., Ramanan, D., Dollár, P., & Zitnick, C. L. (2014) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-10602-1_48 · Cited by 45,460
(2016) · Cited by 31,590
Feature Pyramid Networks for Object Detection
Lin, T.-Y., Dollar, P., Girshick, R., He, K., Hariharan, B., & Belongie, S. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.106 · Cited by 30,939 · cited 3×
“21 Faster-RCNN [32] Resnet-50-FPN [29] 72.”
MethodInfluentialGirshick, R. (2015) · 2015 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.169 · Cited by 28,105
Auto-Encoding Variational Bayes
Kingma, D. P., & Welling, M. (2013) · arXiv · doi:10.48550/arxiv.1312.6114 · Cited by 20,067 · cited 1×
“2, and obtain the class probability distribution for the k-th mixture by using the reparameterization trick [25] of applying Gaussian noise and variance Σp to μ k p [24]:”
MethodThe pascal visual object classes (VOC) challenge
Everingham, M., Van Gool, L., Williams, C. K. I., Winn, J., & Zisserman, A. (2010) · International Journal of Computer Vision · doi:10.1007/s11263-009-0275-4 · Cited by 19,599 · cited 2×
“We demonstrate the benefits of our approach on PASCAL VOC [9] and MS-COCO [30] in a single-stage architecture such as SSD [31], and show generalized performance in a two-stage architecture such as Faster-RCNN [32].”
Method
Research works printed but not linked: 4
- 10.5555/3295222.3295309 · DOI
- 10.48550/arXiv.1405.0312 · DOI · Paper page
- 10.1007/978-3-319-46448-0_2 · DOI · Paper page
- 10.1145/584091.584093 · DOI
Cited by
Sorted by citations
Items 101–136 of 136
Not All Labels Are Equal: Rationalizing The Labeling Costs for Training Object Detection
Elezi, I., Yu, Z., Anandkumar, A., Leal-Taixe, L., & Alvarez, J. M. (2021) · arXiv · Cited by 50
Self-citationBridging the Last Mile in Sim-to-Real Robot Perception via Bayesian Active Learning.Preprint
Feng, J., Lee, J., Durner, M., & Triebel, R. (2021) · arXiv (Cornell University)
Knowledge-Driven Active Learning
Ciravegna, G., Precioso, F., Betti, A., Mottin, K., & Gori, M. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-43412-9_3 · Cited by 5
“Finally, as anticipated above, the Knowledge-driven Active Learning (KAL) strategy can be also employed in the object-detection context where standard uncertainty-based ones are difficult to apply (Choi et al., 2021; Haussmann et al., 2020).”
Evaluating active learning methods for synthetic aperture radar maritime object detection
Sato, J., Raheema, J., & Jaszewski, M. (2022) · Artificial Intelligence and Machine Learning in Defense Applications · doi:10.1117/12.2636037
“7 Although representation-based approaches to active learning are task-independent, these methods have nonetheless demonstrated competitive performance on object detection and image classification.”
AEBSR: Active-Sampling and Energy-Based Single Image Super-Resolution
Jiang, B., Long, K., & Yang, Y.-B. (2022) · Proceedings - International Conference on Image Processing · doi:10.1109/icip46576.2022.9897271
“For example, [5] [6] introduce AL into classification and object detection tasks respectively, which use fewer annotations to obtain higher indicators than random sampling.”
MethodEvaluating Zero-Cost Active Learning for Object Detection
Probst, D., Raza, H., & Rodner, E. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-26236-4_4
Pedestrian Attribute Recognition Method Based on Multi-source Teacher Model Fusion
Ding, Z., & Shang, Y. (2023) · Communications in computer and information science · doi:10.1007/978-981-99-0856-1_2
A Joint Framework based on PU Learning and Active Learning for Object Detection
Li, Y., & Jia, M. (2023) · International Conference on Intelligent Computing and Signal Processing (ICSP) · doi:10.1109/icsp58490.2023.10248659
“Active Learning for Object Detection Active learning can reduce annotation costs by improving the selection of selectively labeled samples, and previous research has demonstrated the effectiveness and necessity of active learning methods[4, 5].”
BackgroundMulti-Level Evidence Active Learning for Water Surface Object Detection
Zhong, W., Han, T., Hu, H., Wang, Y., Li, L., & Zan, P. (2023) · International Conference on Image, Vision and Computing (ICIVC) · doi:10.1109/icivc58118.2023.10270469
“Recently, AL-MDN [31] proposed using a Gaussian mixture model (GMM) to fit training data and calculate the epistemic uncertainty, but the method is not stable.”
MethodEfficient Data Curation Using Active Learning for a Video-Based Fire Detection
Joshi, K., Aziz, A., Dietrich, P., & König, M. (2023) · Proceedings e report · doi:10.36253/10.36253/979-12-215-0289-3.60
Visual object detection model integrating deep active learning
Yudong, C., Li, Z., Chen, L., & Xu, J. Y. (2024) · Insights of automation in manufacturing. · doi:10.59782/iam.v1i1.209
AL-GTD: Deep Active Learning for Gaze Target Detection
Tonini, F., Dall'Asen, N., Vaquero, L., Beyan, C., & Ricci, E. (2024) · ACM Multimedia · doi:10.1145/3664647.3680952 · Cited by 1
Ya, W., Tang, H.-Z., Luo, H., Gao, S., & Xu, Y.-B. (2024) · Laser & Optoelectronics Progress · doi:10.3788/lop240632
Jeong, H., & Oh, G. (2024) · Visual Communications and Image Processing · doi:10.1109/vcip63160.2024.10849827
Uncertainty-Based Active Learning for Underwater Image Annotation
Shahlan, N. A., Kahar, N. F., & Ngadiran, R. (2025) · Studies in computational intelligence · doi:10.1007/978-3-031-78918-2_8
On the Two-fold Role of Logic Constraints in Deep Learning
Ciravegna, G. (2025) · Firenze University Press eBooks · doi:10.36253/979-12-215-0680-8
HeAL3D: Heuristical-enhanced Active Learning for 3D Object Detection
Rivera, E., Prabhakaran, S., & Lienkamp, M. (2025) · 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw67362.2025.00006
“Choi et al. [7] introduce a mixture density network to generate a probabilistic distribution for both the classification and localization heads, allowing for the estimation of both aleatoric and epistemic uncertainties in the data.”
ILD: Image-Level Labels Driven Active Learning Object Detection
Tian, R., Zhang, J., Zhang, Y., Zhang, M., Zhang, Z., Zhang, Y., Li, Y., & Zuo, W. (2025) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2025.3613389
“Choi et al. [2] rely on mixture density networks to estimate the uncertainty for further annotating informative images.”
InfluentialOpen-Set Active Learning for Nucleus Detection From the Histopathological Images
Tang, J., Yue, Y., Chu, W., Wang, M., Wang, Y., Wan, P., Krylov, A. S., Shao, W., & Zhang, D. (2025) · IEEE Transactions on Medical Imaging · doi:10.1109/tmi.2025.3617073 · Cited by 1
Active Learning and the Various Flavors of Supervision for Object Detection
Bischoff, N., & Tomforde, S. (2025) · Communications in computer and information science · doi:10.1007/978-3-032-01169-5_19
Optimizing Active Learning in Vision-Language Models via Parameter-Efficient Uncertainty Calibration
Narayanan, A., Machireddy, A., & Krishnan, R. (2025) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn64981.2025.11228515
“In uncertainty-based sampling, the samples are chosen based on predicted uncertainty metrics such as entropy [24], mixture density networks [25] and margin [26] to rank the samples.”
Yu, C., Zeng, H., Liu, T., & Li, J. (2025) · IEEE International Conference on Robotics and Biomimetics (ROBIO) · doi:10.1109/robio66223.2025.11378229
“Object recognition [2] is a fundamental task that generally falls within the broader domain of object detection. which includes both traditional vision-based[3] methods and deep learning-based algorithms.”
OmViD: Omni-Supervised Active Learning for Video Action Detection
Rana, A., Kumar, A., Vineet, V., & Rawat, Y. (2025) · 2025 IEEE/CVF International Conference on Computer Vision Workshops (ICCVW) · doi:10.1109/iccvw69036.2025.00720 · Cited by 3
“There has been wide study for image based object classification [24, 37] and detection [9, 56, 78] to find a new subset of data for annotation, often using an oracle.”
Exploiting Highly Inconsistent Image Regions for Active Semi-Supervised Crowd Counting
Xu, S., Dong, H., & Zhang, C. (2026) · IEEE Signal Processing Letters · doi:10.1109/lsp.2026.3676320
“Moreover, many AL approaches have been applied in different fields of computer vision, such as image classification [18], [19], [20] and object detection [21], [22], [23].”
Han, T., Wang, J., Zhao, Y., Lin, Y., Yu, L., He, S., Zan, P., & Bo, X. (2026) · Molecules · doi:10.3390/molecules31071144
Annotation-efficient learning for object discovery and detection
Vo, H. V. (2022) · theses.fr (ABES) · doi:10.70675/aae8cd95z8e92z4d35z9384zafe56efdad87
Frequency-Guided Multi-Level Reasoning for Scene Graph Generation in Video
Li, C., Duan, Y., & Tao, X. (2026) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp55912.2026.11464376
“Unlike traditional GMM [23], GMM-Plus introduces a frequency-aware regularization term to capture diverse visual modes .”
Multi-Instance Active Learning with Spatial-Aware Attention for UAVs Object Detection
Qi, X., Wang, H.-P., Zong, J., & Luo, H.-Q. (2026) · Chinese Control and Decision Conference · doi:10.1109/ccdc69976.2026.11560800
“ALMDN [7] leverages a Gaussian Mixture Model to estimate uncertainties in a unified manner for object detection tasks.”
Hwang, S., Kim, S., Jeong, H., & Kum, D. (2026) · 2026 IEEE Intelligent Vehicles Symposium (IV) · doi:10.1109/iv66570.2026.11623840
Activation-Aware Active Learning for Efficient Dataset Curation
Olber, B., Łyskawa, J., & Radlak, K. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-35506-5_58
Sensor Placement with Neural Processes by Mixture Component Disagreement
Eksen, F., Oehmcke, S., & Lüdtke, S. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37667-1_18
Lee, M.-S., & Han, S.-Y. (2026) · Systems · doi:10.3390/systems14091126 · Cited by 1
Le, V.-D.-K., Ngo, V.-T.-D., Tran, T. V., & Nguyen, D.-N.-Q. (2026) · International Conference on Multimedia Analysis and Pattern Recognition (MAPR) · doi:10.1109/mapr72750.2026.11685724
“Although AL and HITL have shown promising results in computer vision [12]–[14], their integration into real-time embedded waste sorting systems remains limited.”
Temporal Consistency-Based Data Selection for Active Learning in Construction Site Object Detection
Lee, S. H., Lee, J., Jeong, J. W., & Ahn, S. (2026) · Journal of Construction Automation and Robotics · doi:10.55785/jcar.5.3.8
FPAL: a full-process probability-driven active learning for whole slide image diagnosis
Xu, Z., Cai, Y., Han, T., Jia, X., Xu, W., & Yang, N. (2026) · Pattern Analysis and Applications · doi:10.1007/s10044-026-01792-y
“Inspired by previous research [11], to select the most informative samples from the unlabeled pool, specifically, those that exhibit high inherent noise and pose significant chal - lenges to the model for accurate classification, an uncer - tainty query strategy based on multi-component probability…”
Mutz, F. W., Silva, F. B. C., Badue, C. S., de Souza, A. F., & Oliveira-Santos, T. (2026) · IEEE Sensors Applications Symposium (SAS) · doi:10.1109/sas70013.2026.11744126
“Models can be overconfident on rare or out-of-distribution samples, while assigning high uncertainty to common scenes affected by factors such as illumination or motion blur [11], [15], [16].”
Metadata
- Authors
- Choi, J., Elezi, I., Lee, H.-J., Farabet, C., & Alvarez, J. M.
- Affiliations
- Seoul National University (KR)
- Ca' Foscari University of Venice (IT)
- NVIDIA
- Venue
- 2021 IEEE/CVF International Conference on Computer Vision (ICCV)
- Publisher
- IEEE
- Published
- 2021-10
- Pages
- 10244-10253
- OpenAlex
- W3145121851
- Cited by
- 135
- References
- 49
- Citations by year
- 2026: 15 · 2025: 27 · 2024: 36 · 2023: 36 · 2022: 17 · 2021: 4
- FWCI
- 10.74
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- IEEE/CVF International Conference on Computer Vision (ICCV) (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning and Algorithms
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Physical Sciences › Computer Science › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
- Keywords
- active learning (machine learning), object detection, mixture density networks, uncertainty estimation, aleatoric uncertainty, epistemic uncertainty, annotation cost reduction
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
