The DOI matches its Crossref record; 66 of the 82 references it cites were found in a registry or an index.
Categorical Depth Distribution Network for Monocular 3D Object Detection
Open accessAbstract
Monocular 3D object detection is a key problem for autonomous vehicles, as it provides a solution with simple configuration compared to typical multi-sensor systems. The main challenge in monocular 3D detection lies in accurately predicting object depth, which must be inferred from object and scene cues due to the lack of direct range measurement. Many methods attempt to directly estimate depth to assist in 3D detection, but show limited performance as a result of depth inaccuracy. Our proposed solution, Categorical Depth Distribution Network (CaDDN), uses a predicted categorical depth distribution for each pixel to project rich contextual feature information to the appropriate depth interval in 3D space. We then use the computationally efficient bird’s-eye-view projection and single-stage detector to produce the final output detections. We design CaDDN as a fully differentiable end-to-end approach for joint depth estimation and object detection. We validate our approach on the KITTI 3D object detection benchmark, where we rank 1stamong published monocular methods. We also provide the first monocular 3D detection results on the newly released Waymo Open Dataset. We provide a code release for CaDDN which is made available.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
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?
In the citing paper
“A popular BEV framework that generates BEV features is based on depth information [46, 32, 34], but this paradigm is sensitive to the accuracy of depth values or the depth distributions.”
From BEVFormer: Learning Bird's-Eye-View Representation from Multi-Camera Images via Spatiotemporal Transformers (2022) · cites this paper as [34]
In the citing paper
“BEV-based approaches have not shown significant advantages over other paradigm in 3D object detections [47, 31, 34].”
From BEVFormer: Learning Bird's-Eye-View Representation from Multi-Camera Images via Spatiotemporal Transformers (2022) · cites this paper as [34]
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
Adam: A Method for Stochastic Optimization
Kingma, D. P., & Ba, J. (2014) · arXiv · doi:10.48550/arxiv.1412.6980 · Cited by 98,946
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
Fully convolutional networks for semantic segmentation
Long, J., Shelhamer, E., & Darrell, T. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298965 · Cited by 38,834
PyTorch: An Imperative Style, High-Performance Deep Learning Library
Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., Killeen, T., Lin, Z., Gimelshein, N., Antiga, L., et al. (2019) · arXiv · doi:10.48550/arxiv.1912.01703 · Cited by 31,473
Focal Loss for Dense Object Detection
Lin, T.-Y., Goyal, P., Girshick, R., He, K., & Dollar, P. (2017) · 2017 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2017.324 · Cited by 30,581
Chen, L.-C., Papandreou, G., Kokkinos, I., Murphy, K., & Yuille, A. L. (2018) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2017.2699184 · Cited by 22,392
Chen, L.-C., Papandreou, G., Kokkinos, I., Murphy, K., & Yuille, A. L. (2016) · arXiv · doi:10.48550/arxiv.1606.00915 · Cited by 22,392
Are we ready for autonomous driving? The KITTI vision benchmark suite
Geiger, A., Lenz, P., & Urtasun, R. (2012) · 2012 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2012.6248074 · Cited by 14,897
Focal Loss for Dense Object Detection
Lin, T.-Y., Goyal, P., Girshick, R., He, K., & Dollar, P. (2020) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2018.2858826 · Cited by 9,556
Flagged first, then by citations
16 of the 82 references couldn’t be matched to a known paper.
Research works printed but not linked: 4
- mani, 2020, Monolayout: Amodal scene layout from a single image, WACV
- ma, 2020, Rethinking pseudolidar representation, ECC
- shi, 2020, Distance-normalized unified representation for monocular 3d object detection, ECC
- roddick, 2018, Orthographic feature transform for monocular 3D object detection, BMV
Cited by
Sorted by citations
Items 101–200 of 597
Detecting as Labeling: Rethinking LiDAR-Camera Fusion in 3D Object Detection
Huang, J., Ye, Y., Liang, Z., Shan, Y., & Du, D. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-72670-5_25 · Cited by 47
DSGN++: Exploiting Visual-Spatial Relation for Stereo-Based 3D Detectors
Chen, Y., Huang, S., Liu, S., Yu, B., & Jia, J. (2022) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2022.3197236 · Cited by 45
GitNet: Geometric Prior-Based Transformation for Birds-Eye-View Segmentation
Gong, S., Ye, X., Tan, X., Wang, J., Ding, E., Zhou, Y., & Bai, X. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-19769-7_23 · Cited by 39
Li, Z., Wang, W., Li, H., Xie, E., Sima, C., Lu, T., Yu, Q., & Dai, J. (2022) · arXiv · doi:10.48550/arxiv.2203.17270 · Cited by 1,419
“BEVFormer outperforms CaDNN with APH of 13.3% and 11.2% on LEVEL_1 and LEVEL_2 difficulties with IoU criteria of 0.5.”
BackgroundMethodInfluentialOPA-3D: Occlusion-Aware Pixel-Wise Aggregation for Monocular 3D Object Detection
Su, Y., Di, Y., Zhai, G., Manhardt, F., Rambach, J. R., Busam, B., Stricker, D., & Tombari, F. (2023) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2023.3238137 · Cited by 38
PPF-Det: Point-Pixel Fusion for Multi-Modal 3D Object Detection
Xie, G., Zhi-yuan, C., Gao, M., Hu, M., & Qin, X. (2024) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2023.3347078 · Cited by 39
“Currently, 3D object detection methods can be roughly divided into LiDAR-based [2], [3], [4], [5], [6], [7], [8], camera-based [9], [10], [11], [12], [13], [14], [15], and camera-LiDAR based [1], [16], [17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27].”
MethodSong, R., Liang, C., Cao, H., Yan, Z., Zimmer, W. M. X., Gross, M., Festag, A., & Knoll, A. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.01704 · Cited by 45
“Following the work in [11, 12], we employ CaDNN [3] for depth estimation.”
MethodMultivariate Probabilistic Monocular 3D Object Detection
Shi, X., Chen, Z., & Kim, T. (2023) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv56688.2023.00426 · Cited by 39
“BEVbased methods [40, 38] first transform the feature maps from perspective view to orthographic view, then directly conduct 3D object detection in the 3D space.”
MethodExploring Geometric Consistency for Monocular 3D Object Detection
Lian, Q., Ye, B., Xu, R., Yao, W., & Zhang, T. (2022) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52688.2022.00173 · Cited by 48
SupFusion: Supervised LiDAR-Camera Fusion for 3D Object Detection
Qin, Y., Wang, C., Kang, Z., Ma, N., Li, Z., & Zhang, R. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.02012 · Cited by 40
“Many methods are proposed to address monocular 3D detection[24, 23, 17, 51, 53, 27, 37, 36] since KITTI benchmark[9] only contains one from the camera.”
MethodAerial Monocular 3D Object Detection
Hu, Y., Fang, S., Xie, W., & Chen, S. (2023) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2023.3245421 · Cited by 36
“Recently, [15], [16] jointly performs depth estimation and detection, and uses the estimated depth to weight the contribution of image features to the grids.”
BackgroundBEVRefiner: Improving 3D Object Detection in Bird’s-Eye-View via Dual Refinement
Wang, B., Zheng, H., Zhang, L., Liu, N., Anwer, R. M., Cholakkal, H., Zhao, Y., & Li, Z. (2024) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2024.3394550 · Cited by 36
“As of to date, there exist two paradigms for multi-view camera based 3D detection that operate in 3D space: depth-based [1], [2], [5], [9], [26], [27], [28] and attention-based [4], [29], [30], [31], [32], [33], [34] methods.”
BackgroundLearning Occupancy for Monocular 3D Object Detection
Peng, L., Xu, J., Cheng, H., Yang, Z., Wu, X., Qian, W., Wang, W., Wu, B., & Cai, D. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.00979 · Cited by 38
“The most related work to ours is CaDDN [53].”
BackgroundMethodInfluentialMonoDGP: Monocular 3D Object Detection with Decoupled-Query and Geometry-Error Priors
Pu, F., Wang, Y. Q., Deng, J., & Yang, W. (2025) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52734.2025.00611 · Cited by 35
Boosting 3D Object Detection by Simulating Multimodality on Point Clouds
Wu, Z., Hong, M., Jiang, L., & Fu, C. (2022) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52688.2022.01327 · Cited by 35
“On the other hand, camera-produced images are a popular modality for monocular and stereo 3D object detection [7,16,21,22,31].”
BackgroundDeepFusion: A Robust and Modular 3D Object Detector for Lidars, Cameras and Radars
Drews, F., Feng, D., Faion, F., Rosenbaum, L., Ulrich, M., & Gläser, C. (2022) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros47612.2022.9981778 · Cited by 39
“On the other hand, radars and monocular cameras suffer from sparse or imprecise depth estimation [5]–[7], which can be compensated by”
BackgroundMonoLSS: Learnable Sample Selection For Monocular 3D Detection
Li, Z., Jia, J., & Shi, Y. (2024) · International Conference on 3D Vision (3DV) · doi:10.1109/3dv62453.2024.00088 · Cited by 40
“Although MonoLSS’s performance is slightly lower than that of CaDDN [42], it is worth noting that CaDDN relies on LiDAR data during training, whereas MonoLSS is an image-only approach.”
BackgroundMethodCam4DOcc: Benchmark for Camera-Only 4D Occupancy Forecasting in Autonomous Driving Applications
Ma, J., Chen, X., Huang, J., Xu, J., Luo, Z., Xu, J., Gu, W., Ai, R., & Wang, H. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.02030 · Cited by 34
“Traditional camera-based perception methods for object detection [1], [2], [3], [4], semantic segmentation [5], [6], [7], [8], and panoptic segmentation [9], [10], [11], [12] focus on predefined specific object categories, making them less effective at recognizing uncommon objects.”
BackgroundLiDAR-camera fusion: Dual transformer enhancement for 3D object detection
Chen, M., Liu, P., & Zhao, H. (2023) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2022.105815 · Cited by 34
PDR: Progressive Depth Regularization for Monocular 3D Object Detection
Hua, X.-S., Cai, S., Zhao, N., Deng, B., Zhao, M., & Lee, G. H. (2023) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2023.3276518 · Cited by 33
“Many existing works [13], [14], [15], [16], [17], [18], [19] rely on extra pixel-level depth annotations to construct a depth estimator for monocular 3D detection.”
BackgroundMethodSparseLIF: High-Performance Sparse LiDAR-Camera Fusion for 3D Object Detection
Zhang, H., Liang, L., Zeng, P., Song, X., & Wang, Z. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-72761-0_7 · Cited by 35
Ray denoising: Depth-aware hard negative sampling for multi-view 3d object detection
Liu, F., Huang, T., Zhang, Q., Yao, H., Zhang, C., Wan, F., Ye, Q., & Zhou, Y. (2025) · Lecture Notes in Computer Science · doi:10.1007/978-3-031-72967-6_12 · Cited by 33
SSD-MonoDETR: Supervised Scale-Aware Deformable Transformer for Monocular 3D Object Detection
He, X., Yang, F., Yang, K., Lin, J., Fu, H., Wang, M., Yuan, J., & Li, Z. (2023) · IEEE Transactions on Intelligent Vehicles · doi:10.1109/tiv.2023.3311949 · Cited by 33
“To boost the performance, many approaches explore the assistance of external resources like depth maps [20], [28], [29], [30] or LiDAR [26], [31], [32].”
BackgroundMonoPGC: Monocular 3D Object Detection with Pixel Geometry Contexts
Wu, Z., Gan, Y., Wang, L., Chen, G., & Pu, J. (2023) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48891.2023.10161442 · Cited by 32
“Following previous methods [1], our approach is compared on the test set using AP40 instead of AP11.”
BackgroundMethodInfluentialJiang, H., Lu, Y., Zhang, D., Shi, Y., & Wang, J. (2024) · Applied Soft Computing · doi:10.1016/j.asoc.2024.111253 · Cited by 33
A LiDAR–Camera Fusion 3D Object Detection Algorithm
Liu, L., He, J. Y., Ren, K., Xiao, Z., & Hou, Y. (2022) · Information · doi:10.3390/info13040169 · Cited by 34
“CaDDN [9] generated depth distributions for each pixel of the image and combined them with image features for camera-based 3D object detection.”
D-IF: Uncertainty-aware Human Digitization via Implicit Distribution Field
Yang, X., Luo, Y., Xiu, Y., Wang, W., Xu, H., & Fan, Z. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.00837 · Cited by 35
“Additionally, CaDDN [34] projects rich contextual features into the appropriate depth interval in 3D space using a projected categorical depth distribution for each pixel.”
BackgroundCalibration-Free BEV Representation for Infrastructure Perception
Fan, S., Wang, Z., Huo, X., Wang, Y., & Liu, J. (2023) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros55552.2023.10341916 · Cited by 30
“Considering feature-level fusion, feature lifting-based methods [10], [11], [12], [13], [14], [15] first transform 2D image feature into 3D voxel feature and collapse them to generate BEV features.”
MethodMonoDiff: Monocular 3D Object Detection and Pose Estimation with Diffusion Models
Ranasinghe, Y., Hegde, D., & Patel, V. M. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.01014 · Cited by 32
“We adopt the performance reporting criterion of CaDDN [58] for a fair comparison with existing methods.”
MethodPV-RCNN++: semantical point-voxel feature interaction for 3D object detection
Wu, P., Gu, L., Yan, X., Xie, H., Wang, F. L., Cheng, G., & Wei, M. (2022) · The Visual Computer · doi:10.1007/s00371-022-02672-2 · Cited by 34
“Much progress has been made in 3D object detection via various data representation (e.g., monocular images Yan & Salman (2017); Reading et al. (2021); Chen et al. (2020); Li et al. (2019a), stereo cameras Chen et al. (2017a); Li et al. (2019b), and LiDAR point clouds).”
BackgroundMethodInfluentialUnsupervised Domain Adaptation for Monocular 3D Object Detection via Self-training
Li, Z., Chen, Z., Li, A., Fang, L., Jiang, Q., Liu, X., & Jiang, J. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-20077-9_15 · Cited by 30
TransCAR: Transformer-Based Camera-and-Radar Fusion for 3D Object Detection
Pang, S., Morris, D., & Radha, H. (2023) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros55552.2023.10341793 · Cited by 36
Occlusion-Aware Plane-Constraints for Monocular 3D Object Detection
Yao, H., Chen, J., Wang, Z., Wang, X., Han, P., Chai, X., & Qiu, Y. (2023) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2023.3323036 · Cited by 29
Collaborative Multimodal Fusion Network for Multiagent Perception
Zhang, L., Wang, B., Zhao, Y., Yuan, Y., Zhou, T., & Li, Z. (2024) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2024.3491756 · Cited by 32
“D image backbone for visual encoding and an LSS-like [62], [63] architecture for explicit view transformation.”
DevNet: Self-supervised Monocular Depth Learning via Density Volume Construction
Zhou, K., Hong, L., Chen, C., Xu, H., Ye, C., Hu, Q., & Li, Z. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-19842-7_8 · Cited by 30
Shape-Aware Monocular 3D Object Detection
Chen, W., Zhao, J., Zhao, W., & Wu, S.-Y. (2023) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2023.3249909 · Cited by 29
AeDet: Azimuth-Invariant Multi-View 3D Object Detection
Feng, C., Jie, Z., Zhong, Y., Chu, X., & Ma, L. (2023) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52729.2023.02067 · Cited by 29
“[22] projects the image feature to the efficient BEV feature to perform 3D object detection.”
BackgroundBEVSOC: Self-Supervised Contrastive Learning for Calibration-Free BEV 3-D Object Detection
Chen, Y., Li, N., Zhu, D., Zhou, C., Hu, Z., Bai, Y., & Yan, J. (2024) · IEEE Internet of Things Journal · doi:10.1109/jiot.2024.3379471 · Cited by 29
NeurOCS: Neural NOCS Supervision for Monocular 3D Object Localization
Min, Z., Zhuang, B., Schulter, S., Liu, B., Dunn, E., & Chandraker, M. K. (2023) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52729.2023.02050 · Cited by 28
“In light of depth being the most critical factor [25, 43], many works develop depth-guided localization, including depth-aware networks [3, 13, 24, 28], depth-conditioned message propagation [67], depth-equivariant network [28], depth-guided feature projection [58], depth-based feature enhancement [1], depth pretraini…”
BackgroundExploring Point-BEV Fusion for 3D Point Cloud Object Tracking With Transformer
Luo, Z., Zhou, C., Pan, L., Zhang, G., Liu, T., Luo, Y., Zhao, H., Liu, Z., & Lu, S. (2024) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2024.3373693 · Cited by 28
Lite-FPN for keypoint-based monocular 3D object detection
Yang, L., Zhang, X., Li, J., Wang, L., Zhu, M., & Zhu, L. (2023) · Knowledge-Based Systems · doi:10.1016/j.knosys.2023.110517 · Cited by 33
A Simple Baseline for Multi-Camera 3D Object Detection
Zhang, Y., Zheng, W., Zhu, Z. D., Huang, G., Lu, J., & Zhou, J. (2023) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v37i3.25460 · Cited by 27
Multi-Sem Fusion: Multimodal Semantic Fusion for 3-D Object Detection
Xu, S., Li, F., Song, Z., Fang, J., Wang, S., & Yang, Z.-X. (2024) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2024.3387732 · Cited by 26
“In previous image-based 3D object detection methods, features are extracted by constructing a network from either single or multiple images for predicting 3D bounding boxes.”
MethodZhu, M., Gong, Y., Tian, C., & Zhu, Z. (2024) · Drones · doi:10.3390/drones8080412 · Cited by 30
“The first category involved using pre-trained monocular depth estimation models, demonstrated by techniques like Pseudo-LiDAR [52], and CADDN [53].”
MethodTwo-branch Objectness-centric Open World Detection
Wu, Y. X., Zhao, X., Ma, Y., Wang, D., & Liu, X. (2022) · International Workshop on Human-centric Multimedia Analysis (HCMA) · doi:10.1145/3552458.3556453 · Cited by 26
“Along with the development of powerful convolutional neural networks (CNNs) [8, 12] and large-scale well-annotated datasets, the performance of object detection networks [3, 13, 16, 22, 23] has achieved remarkable improvement, and benefits a wide range of applications [6, 9, 15].”
BackgroundToward Collaborative Autonomous Driving: Simulation Platform and End-to-End System
Liu, G., Hu, Y., Xu, C., Mao, W., Ge, J., Huang, Z., Lu, Y., Xu, Y., Xia, J., Wang, Y., et al. (2025) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2025.3560327 · Cited by 39
MonoCInIS: Camera Independent Monocular 3D Object Detection using Instance Segmentation
Heylen, J., Wolf, M. D., Dawagne, B., Proesmans, M., Gool, L. V., Abbeloos, W., Abdelkawy, H., & Reino, D. O. (2021) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw54120.2021.00108 · Cited by 25
“CaDDN[22] uses categorical depth distribution for each pixel to project contextual feature information to the appropriate depth interval in 3D space, and then uses a BEV projection to produce the final output bounding boxes.”
One Stage Monocular 3D Object Detection Utilizing Discrete Depth and Orientation Representation
Haq, M. A., Ruan, S.-J., Shao, M.-E., Haq, Q. M. U., Liang, P.-J., & Gao, D.-Q. (2022) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2022.3175198 · Cited by 25
“The deep estimation network learns to estimate the target object’s distance using categorical depth distribution inspired by CADDN [22].”
BackgroundMethodMonoEdge: Monocular 3D Object Detection Using Local Perspectives
Zhu, M., Ge, L., Wang, P., & Peng, H. (2023) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv56688.2023.00071 · Cited by 27
Wu, Y., Li, R., Qin, Z., Zhao, X., & Li, X. (2024) · IEEE Transactions on Image Processing · doi:10.1109/tip.2024.3427701 · Cited by 33
“Following LSS, CaDDN [18] and BEVDet [19] utilize the depth distribution to construct the BEV representation.”
MethodWang, W., Lu, Y., Zheng, G., Zhan, S., Ye, X., Tan, Z., Wang, J., Wang, G., & Li, X. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.01394 · Cited by 29
“Extensive work follows this setting [14, 16, 25, 44].”
QE-BEV: Query Evolution for Bird's Eye View Object Detection in Varied Contexts
Yao, J., Lai, Y., Kou, H., Wu, T., & Liu, R. (2024) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3664647.3680807 · Cited by 25
“Monocular 3D object detection [24, 30, 31] and multiview approaches [9, 23] have been widely studied for generating 3D bounding boxes from 2D images.”
SimDistill: Simulated Multi-Modal Distillation for BEV 3D Object Detection
Zhao, H., Zhang, Q., Zhao, S., Chen, Z., Zhang, J., & Tao, D. (2024) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v38i7.28577 · Cited by 28
“Camera-based 3D Object Detection Monocular 3D object detection methods have been widely studied and made great progress (Simonelli et al. 2019; Reading et al. 2021; Wang et al. 2021b; Lu et al. 2021; Ma et al. 2021; Huang et al. 2022a) on the KITTI (Geiger, Lenz, and Urtasun 2012) benchmark.”
MethodWayFASTER: a Self-Supervised Traversability Prediction for Increased Navigation Awareness
Gasparino, M. V., Sivakumar, A. N. V., & Chowdhary, G. V. (2024) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra57147.2024.10610436 · Cited by 30
“The visual depth prediction is trained as a multi-class classification problem, where each depth discretized bin is treated as a class [19].”
BackgroundBEV-Locator: an end-to-end visual semantic localization network using multi-view images
Zhang, Z., Xu, M., Zhou, W., Peng, T., Li, L., & Poslad, S. (2025) · Science China Information Sciences · doi:10.1007/s11432-023-4114-6 · Cited by 24
“Following work includes BEV-Seg [43], CaDDN [30], FIERY [40], BEVDet [31], and BEVDepth”
Unleashing HyDRa: Hybrid Fusion, Depth Consistency and Radar for Unified 3D Perception
Wolters, P., Gilg, J., Teepe, T., Herzog, F., Laouichi, A., Hofmann, M. R., & Rigoll, G. (2025) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra55743.2025.11127412 · Cited by 23
“Shifting from perspective-view 3D object detection [50, 59, 68, 69, 83] to Birds-Eye-View (BEV) based perception [56, 57], the BEV-based architectures have become the new standard for 3D object detection and tracking [28].”
Wu, Z., Wu, Y., Pu, J. Y., Li, X., & Wang, X. (2023) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v37i3.25391 · Cited by 23
“While performance gains on PatchNet (Ma et al. 2020) are not as obvious as those for MonoDETR (Zhang et al. 2022) and CaDDN (Reading et al. 2021) (Fu et al. 2018), are as competitive as GT depth from direct LiDAR projection, and thus retrains the effectiveness of guiding monocular student with…”
BackgroundMethodInfluentialSelective Transfer Learning of Cross-Modality Distillation for Monocular 3D Object Detection
Ding, R., Yang, M., & Zheng, N. (2024) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2024.3405992 · Cited by 22
UniMODE: Unified Monocular 3D Object Detection
Li, Z., Xu, X., Lim, S.-N., & Zhao, H. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.01567 · Cited by 22
“1 is similar to the linear-increasing discretization of depth bin in CaDDN [26], while the essence is fundamentally different.”
BackgroundFD3D: Exploiting Foreground Depth Map for Feature-Supervised Monocular 3D Object Detection
Wu, Z., Gan, Y., Wu, Y., Wang, R., Wang, X., & Pu, J. Y. (2024) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v38i6.28436 · Cited by 22
“For KITTI dataset, following prior works (Reading et al. 2021), the 3D Average Precision ( AP 3 D ) and BEV Average Precision ( AP BEV ) are two vital evaluation metrics.”
BackgroundMethodInfluentialWang, Z., Fan, S., Huo, X., Xu, T., Wang, Y., Liu, J., Chen, Y., & Zhang, Y. (2024) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra57147.2024.10610545 · Cited by 21
“Most methods [23], [24], [25], [26], [27] applied 2D-to-3D transformation following LSS [28], which predicted a depth distribution for each pixel and lifted image features into frustum features with camera parameters, then splatted all frustums into a rasterized BEV feature.”
MethodDimension Embeddings for Monocular 3D Object Detection
Zhang, Y., Zheng, W., Zhu, Z. D., Huang, G., Du, D., Zhou, J., & Lu, J. (2022) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52688.2022.00164 · Cited by 22
EV-Planner: Energy-Efficient Robot Navigation via Event-Based Physics-Guided Neuromorphic Planner
Sanyal, S., Manna, R. K., & Roy, K. (2024) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2024.3350982 · Cited by 28
“AsshowninTableI,thereareworksthateither1)es-timatedepthusingnon-depthsensors[10],[24],[25],[26],[31], [32]or2)usedepthsensors[33],[34],[35],[36].”
PARTNER: Level up the Polar Representation for LiDAR 3D Object Detection
Nie, M., Xue, Y., Wang, C., Ye, C., Xu, H., Zhu, X., Huang, Q., Mi, M. B., Wang, X., & Li, Z. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.00352 · Cited by 20
“In addition to uniform partition (UD), two widely used range discretization strategies, i.e ., spacing-increasing discretization [9] (SID) and linear-increasing discretization [22] (LID), are also studied.”
MethodBai, L., Li, Y., Cen, M., & Hu, F. (2021) · Remote Sensing · doi:10.3390/rs13163288 · Cited by 21
“The P-R curve of detection results with different methods (BirdNet [44], CaDDN [45], MonoRCNN [46], YoloMono3D [47], RT3D [48], QT-3DT [49], OURS).”
MethodDepth-Guided Vision Transformer With Normalizing Flows for Monocular 3D Object Detection
Pan, C., Peng, J., & Zhang, Z. (2024) · IEEE/CAA Journal of Automatica Sinica · doi:10.1109/jas.2023.123660 · Cited by 26
LET-3D-AP: Longitudinal Error Tolerant 3D Average Precision for Camera-Only 3D Detection
Hung, W.-C., Casser, V., Kretzschmar, H., Hwang, J.-J., & Anguelov, D. (2024) · 2024 IEEE International Conference on Robotics and Automation (ICRA) · doi:10.1109/icra57147.2024.10609986 · Cited by 33
“2) The second camp leverages mature LiDAR detectors and projects 2D perspective features into 3D in the form of point clouds [16], [20], [35]–[40], feature maps in bird’s eye view (BEV) [15]– [17], [41]–[52], or in voxel space [18], [19], [53], [54].”
BackgroundGroupLane: End-to-End 3D Lane Detection With Channel-Wise Grouping
Li, Z., Han, C., Ge, Z., Yang, J., Yu, E., Wang, H., Zhang, X., & Zhao, H. (2024) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2024.3475881 · Cited by 26
“Utilizing the predicted depth information and camera parameters, they can project image features into the 3D space and splat them as the BEV feature [32].”
BackgroundValverde, M., Moutinho, A., & Zacchi, J.-V. (2025) · Sensors · doi:10.3390/s25175264 · Cited by 29
Boosting Monocular 3D Object Detection With Object-Centric Auxiliary Depth Supervision
Kim, Y.-S., Kim, S., Sim, S., Choi, J. W., & Kum, D. (2022) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2022.3224082 · Cited by 21
“CaDDN [40] trains the network to predict pixel-wise depth distribution using dense depth map and transforms the front view image features to bird’s eye view (BEV) to detect 3D objects on BEV features.”
BackgroundMethodMulti-view LiDAR Guided Monocular 3D Object Detection
Liu, J., & Yu, W. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-18916-6_42 · Cited by 19
OBMO: One Bounding Box Multiple Objects for Monocular 3D Object Detection
Huang, C., He, T., Ren, H., Wang, W., Lin, B., & Cai, D. (2023) · IEEE Transactions on Image Processing · doi:10.1109/tip.2023.3333225 · Cited by 19
“We use the same data processing strategy proposed in CaDDN [23].”
BackgroundMethodInfluentialYu, F., Lian, J., Li, L., & Zhao, J. G. (2024) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2024.108124 · Cited by 21
Lin, H., Liu, Y., Wang, L., & Qu, X. (2025) · IEEE Transactions on Big Data · doi:10.1109/tbdata.2025.3527208 · Cited by 24
“3DbboxmethodsincludeMonoDIS[53],D4LCN[54],RTM3D[55],CaDDN[56],andMonoRUn [57]; BEV-based approaches consist of BEVDet [58], BEVerse [59], BEVDepth [60], BEVFormer [61], StreamPETR [62], and SparseBEV [63]; UniOcc [64], TPVFormer [65], OCC3D [66], RenderOcc [67], and OccTransformer [68] are based on…”
LabelDistill: Label-Guided Cross-Modal Knowledge Distillation for Camera-Based 3D Object Detection
Kim, S., Kim, Y.-S., Hwang, S., Jeong, H., & Kum, D. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-72992-8_2 · Cited by 18
UrbanNet: Leveraging Urban Maps for Long Range 3D Object Detection
Carrillo, J. S., & Waslander, S. L. (2021) · IEEE International Conference on Intelligent Transportation Systems (ITSC) · doi:10.1109/itsc48978.2021.9564840 · Cited by 19
Self-citationJPerceiver: Joint Perception Network for Depth, Pose and Layout Estimation in Driving Scenes
Zhao, H., Zhang, J., Zhang, S., & Tao, D. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-19839-7_41 · Cited by 19
Kurniawan, I. T., & Trilaksono, B. R. (2023) · IEEE Access · doi:10.1109/access.2023.3328953 · Cited by 22
“Recent multiple camera methods [39], [40], [41], [42], [43] extract surround-view features from the multiple images and then transform them into the BEV plane.”
MethodTransformer-Based Optimized Multimodal Fusion for 3D Object Detection in Autonomous Driving
Alaba, S. Y., & Ball, J. E. (2024) · IEEE Access · doi:10.1109/access.2024.3385439 · Cited by 21
“The voxel image features are then constructed from these image features by first generating the depth information of each image like in [40] and [41].”
BackgroundLi, J., He, X., Zhou, C., Cheng, X., Wen, Y., & Zhang, D. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-72775-7_6 · Cited by 29
GUPNet++: Geometry Uncertainty Propagation Network for Monocular 3D Object Detection
Lu, Y., Ma, X., Yang, L., Zhang, T., Liu, Y., Chu, Q., He, T., Li, Y., & Ouyang, W. (2024) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2024.3475583 · Cited by 18
“CaDNN [47] utilized depth to transfer the feature-level 2D representation into Bird’s Eye View (BEV) features and treated monocular 3D detection as an end-to-end BEV detection task.”
Sensor Fusion Method for Object Detection and Distance Estimation in Assisted Driving Applications
Favelli, S., Xie, M., & Tonoli, A. (2024) · Sensors · doi:10.3390/s24247895 · Cited by 21
Density-Aware and Semantic-Guided Fusion for 3-D Object Detection Using LiDAR-Camera Sensors
Jhong, S.-Y., Chen, Y.-Y., Hsia, C., Wang, Y.-Q., & Lai, C. (2023) · IEEE Sensors Journal · doi:10.1109/jsen.2023.3302314 · Cited by 18
“have proposed the use of image-based methods [2], [3], [4] to lower the cost of such systems.”
MethodDynamic Point-Pixel Feature Alignment for Multimodal 3-D Object Detection
Wang, J., Kong, X., Nishikawa, H., Lian, Q., & Tomiyama, H. (2023) · IEEE Internet of Things Journal · doi:10.1109/jiot.2023.3329884 · Cited by 18
“D object detection [1], [2], [3] using only cameras in order to achieve this task at a lower cost.”
MethodMonocular 3D Object Detection with Bounding Box Denoising in 3D by Perceiver
Liu, X., Zheng, C., Cheng, K., Xue, N., Qi, G.-J., & Wu, T. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.00592 · Cited by 19
“The problem of monocular 3D object detection is considered ill-posed, prompting recent research efforts to leverage additional information sources, including LiDAR point clouds [4, 10, 19, 28, 45], depth estimation [12, 36, 42, 48, 53, 54, 62], CAD models [25, 34], temporal frames [2], etc.”
BackgroundMethodInfluentialSOGDet: Semantic-Occupancy Guided Multi-View 3D Object Detection
Zhou, Q., Cao, J., Leng, H., Yin, Y., Kun, Y., & Zimmermann, R. (2024) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v38i7.28600 · Cited by 17
Zhao, L., Xu, X., Wang, Z., Zhang, Y., Zhang, B., Zheng, W., Du, D., Zhou, J., & Lu, J. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.00936 · Cited by 17
“Vision-based 3D perception [1, 12, 29, 30, 51] has recently emerged as a promising alternative to LiDAR-based approaches [16, 32, 52], offering an effective means to extract 3D information from 2D images.”
BackgroundM3DGAF: Monocular 3D Object Detection With Geometric Appearance Awareness and Feature Fusion
Chen, M., Liu, P., & Zhao, H. (2022) · IEEE Sensors Journal · doi:10.1109/jsen.2022.3189174 · Cited by 18
MDNet: Multimodal Cooperative Perception via Spatial Alignment of Modal Decision-Making
He, J., Deng, X., Gui, J., Zhang, T., & He, X. (2025) · IEEE Internet of Things Journal · doi:10.1109/jiot.2025.3531145 · Cited by 16
“In this article, we adopt CaDDN [25], a Lift-based method, as the backbone network for camera feature extraction.”
Monocular 3D Object Detection with LiDAR Guided Semi Supervised Active Learning
Hekimoglu, A., Schmidt, M., & Marcos-Ramiro, A. (2023) · IEEE Workshop/Winter Conference on Applications of Computer Vision · doi:10.1109/wacv57701.2024.00234 · Cited by 22
“Following CaDDN [32], we downsample the original training set by selecting every third frame, resulting in a training set of approximately 51K samples labeled with 3D bounding boxes.”
MethodMulti-task Learning for Real-time Autonomous Driving Leveraging Task-adaptive Attention Generator
Choi, W., Shin, M., Lee, H., Cho, J. H., Park, J. H., & Im, S. (2024) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra57147.2024.10610716 · Cited by 16
“The second category compensates for the absence of 3D information in monocular images by utilizing additional data like pre-trained models [28], [29], [30], [31], [32], [33] or CAD models [34], [35].”
Background3D Detection and Pose Estimation of Vehicle in Cooperative Vehicle Infrastructure System
Guo, E., Chen, Z., Rahardja, S., & Yang, J. (2021) · IEEE Sensors Journal · doi:10.1109/jsen.2021.3101497 · Cited by 19
Anytime-Lidar: Deadline-aware 3D Object Detection
Soyyigit, A., Yao, S., & Yun, H. (2022) · IEEE International Conference on Embedded and Real-Time Computing Systems and Applications (RTCSA) · doi:10.1109/rtcsa55878.2022.00010 · Cited by 16
“Afterwards, the generated bird-eye-view pseudo-image is processed by the backbone, which is usually Region Proposal Network (RPN) [8] as can be seen in [7], [9], [14]–[18].”
MethodSaif, A. F. M. S., & Mahayuddin, Z. R. (2022) · International Journal of Advanced Computer Science and Applications · doi:10.14569/ijacsa.2022.0131123 · Cited by 16
“Research in [42] preferred image features in ResNet-101 [4, 17, 43] for block 1 to maintain a high spatial resolution and avoid redundancy of same features.”
BackgroundMethodInfluentialDeepInteraction++: Multi-Modality Interaction for Autonomous Driving
Yang, Z., Song, N., Li, W., Zhu, X., Zhang, L., & Torr, P. H. S. (2025) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2025.3565194 · Cited by 25
Delving Into the Secrets of BEV 3D Object Detection in Autonomous Driving: A Comprehensive Survey
Li, H., Zhao, Y., Zhong, J., Wang, B., Sun, C., & Sun, F. (2025) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2025.3624830 · Cited by 22
“Techniques such as LSS [86] predict depth for each pixel, creating 3D representations, while CaDDN [87] uses ground truth depth supervision to improve depth prediction.”
CrossDTR: Cross-view and Depth-guided Transformers for 3D Object Detection
Tseng, C.-Y., Chen, Y.-R., Lee, H.-Y., Wu, T.-H., Chen, W.-C., & Hsu, W. H. (2023) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48891.2023.10161451 · Cited by 15
“III-C, we borrow the depth-guided method from MonoDETR [31] and refer to CaDDN [10] and adopt Depth Distribution Network Loss (DDN Loss) to regularize the predicted depth map values and predicted depth map logits.”
MethodInfluentialMonocular 3D Object Detection Utilizing Auxiliary Learning With Deformable Convolution
Chen, J.-H., Shieh, J.-L., Haq, M. A., & Ruan, S.-J. (2023) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2023.3319556 · Cited by 15
“Previous methods, such as MonoFENet [69], CaDDN [52], DD3D [55], M3D-RPN [36], MonoFlex [41], and MonoDTR [44], typically necessitate additional feature extraction and reorganization through multiple sub-modules or require supplementary data, such as point clouds or depth information.”
MethodImGeoNet: Image-induced Geometry-aware Voxel Representation for Multi-view 3D Object Detection
Tu, T., Chuang, S.-P., Liu, Y.-L., Sun, C., Zhang, K., Roy, D., Kuo, C.-H., & Sun, M. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.00644 · Cited by 15
“To mitigate the depth information loss, some approaches [52, 13, 47, 68, 74] use an additional backbone for depth map feature extraction, and other approaches [37, 71, 72] back-project depth images to 3D pseudo point clouds.”
MapPrior: Bird’s-Eye View Map Layout Estimation with Generative Models
Zhu, X., Zyrianov, V., Liu, Z., & Wang, S. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.00756 · Cited by 17
“These approaches typically take LiDAR [80, 38, 51, 67], multi-view camera [60, 87, 56, 76, 11, 46, 26, 59, 21], or both [81, 75, 49, 2, 57] as input, and output segmentations for road elements and detections for dynamic traffic participants like cars, cyclists or pedestrians.”
Background
Metadata
- Authors
- Reading, C., Harakeh, A., Chae, J., & Waslander, S. L.
- Affiliations
- University of Toronto (CA)
- Venue
- 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
- Publisher
- IEEE
- Published
- 2021-06
- Pages
- 8551-8560
- ISSN
- 1063-6919 (ISSN-L)
- OpenAlex
- W3133598367
- Cited by
- 595
- References
- 82
- Citations by year
- 2026: 57 · 2025: 143 · 2024: 173 · 2023: 136 · 2022: 73 · 2021: 13
- FWCI
- 32.30
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Licence
- Locations
- IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings (publishedVersion)
- arXiv (Cornell University) (submittedVersion, cc-by)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University) (cc-by)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Physical Sciences › Engineering › Automotive Engineering › Autonomous Vehicle Technology and Safety
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Vision and Imaging
- Keywords
- monocular 3D object detection, depth estimation, single-stage detectors, autonomous driving, Waymo Open Dataset, KITTI dataset
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
