Filters








581 Hits in 3.3 sec

IoU Loss for 2D/3D Object Detection [article]

Dingfu Zhou, Jin Fang, Xibin Song, Chenye Guan, Junbo Yin, Yuchao Dai, Ruigang Yang
<span title="2019-08-11">2019</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
In 2D/3D object detection task, Intersection-over-Union (IoU) has been widely employed as an evaluation metric to evaluate the performance of different detectors in the testing stage.  ...  To eliminate the performance gap between training and testing, the IoU loss has been introduced for 2D object detection in and .  ...  Conclusion and Future Works In this paper, we have addressed the 2D/3D object detection problem by introducing the IoU loss for two rotated Bboxes.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/1908.03851v1">arXiv:1908.03851v1</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/jvhspf5acnejlgjngyre4xlsbe">fatcat:jvhspf5acnejlgjngyre4xlsbe</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20200902155321/https://arxiv.org/pdf/1908.03851v1.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/a7/df/a7df9d524b89fb0e2133ef8ad9dab6f165d74c3a.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/1908.03851v1" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

A survey of Object Classification and Detection based on 2D/3D data [article]

Xiaoke Shen
<span title="2022-01-22">2022</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
Recently, by using deep neural network based algorithms, object classification, detection and semantic segmentation solutions are significantly improved.  ...  makes the detection task more challenging.  ...  The 3D bounding box encoding methods 3 In the following section we will focus on the 3D only or 2D+3D detection systems.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/1905.12683v2">arXiv:1905.12683v2</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/e5ladbmkzjg53c3o6uxzlup3ky">fatcat:e5ladbmkzjg53c3o6uxzlup3ky</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20220128005830/https://arxiv.org/pdf/1905.12683v2.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/4a/ea/4aea70d624e3d8038b5961d9468ed625d3326ce5.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/1905.12683v2" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

Cross-Modality 3D Object Detection [article]

Ming Zhu, Chao Ma, Pan Ji, Xiaokang Yang
<span title="2020-08-16">2020</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
Within the first stage, we further exploit a joint anchor mechanism that enables the network to utilize 2D-3D classification and regression simultaneously for better proposal generation.  ...  To this end, we present a novel two-stage multi-modal fusion network for 3D object detection, taking both binocular images and raw point clouds as input.  ...  Effects of Reprojection Loss As is shown in Table 2 , the 2D-3D reprojection loss plays the most important role for both 2D object detection and 3D object detection.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2008.10436v1">arXiv:2008.10436v1</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/75ib2qseynbetlvm5diqfrfvny">fatcat:75ib2qseynbetlvm5diqfrfvny</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20201012224402/https://arxiv.org/pdf/2008.10436v1.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2008.10436v1" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

Monocular 3D Object Detection with Pseudo-LiDAR Point Cloud [article]

Xinshuo Weng, Kris Kitani
<span title="2019-08-31">2019</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
Following the pipeline of two-stage 3D detection algorithms, we detect 2D object proposals in the input image and extract a point cloud frustum from the pseudo-LiDAR for each proposal.  ...  Then an oriented 3D bounding box is detected for each frustum.  ...  We show that, from only a single RGB image, the 3D bounding box detection for the car category can be very accurate, even for the challenging faraway objects (e.g., in the 6th row 1st column and 8th row  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/1903.09847v4">arXiv:1903.09847v4</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/vwqdxubedzddnkeuos6eqqgsny">fatcat:vwqdxubedzddnkeuos6eqqgsny</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20200828120847/https://arxiv.org/pdf/1903.09847v4.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/cb/cb/cbcb3b48a87fe424de6fe4f3476367fbf006cbe3.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/1903.09847v4" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

Multi-Modality Task Cascade for 3D Object Detection [article]

Jinhyung Park, Xinshuo Weng, Yunze Man, Kris Kitani
<span title="2021-07-08">2021</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
Point clouds and RGB images are naturally complementary modalities for 3D visual understanding - the former provides sparse but accurate locations of points on objects, while the latter contains dense  ...  Despite this potential for close sensor fusion, many methods train two models in isolation and use simple feature concatenation to represent 3D sensor data.  ...  The second stage 3D refinement loss consists of three parts -the box refinement loss, the multiclass semantic classification loss, and the IoU prediction loss: L rcnn = L box-ref ine + L sem-cls + L iou  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2107.04013v1">arXiv:2107.04013v1</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/xl35vqfipbg2ldcc6ichtkk37m">fatcat:xl35vqfipbg2ldcc6ichtkk37m</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20210713060141/https://arxiv.org/pdf/2107.04013v1.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/a2/06/a206ddfc788362fcb759f8664a10e13b0c432bc5.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2107.04013v1" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

AutoShape: Real-Time Shape-Aware Monocular 3D Object Detection [article]

Zongdai Liu, Dingfu Zhou, Feixiang Lu, Jin Fang, Liangjun Zhang
<span title="2021-08-25">2021</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
Then the 2D/3D geometric constraints are built by these correspondences for each object to boost the detection performance.  ...  In this work, we propose an approach for incorporating the shape-aware 2D/3D constraints into the 3D detection framework.  ...  The key technique is simultaneously optimizing the 2D/3D constraints (loss) for better matching. Here, we conduct an ablation study to justify the effectiveness of the 2D/3D loss.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2108.11127v1">arXiv:2108.11127v1</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/o6j6tgomlbcfvb46sct6krfctq">fatcat:o6j6tgomlbcfvb46sct6krfctq</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20210830050405/https://arxiv.org/pdf/2108.11127v1.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/a5/25/a5253b176ac3b00c2596298b37038be6e9f7ee78.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2108.11127v1" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

Lite-FPN for Keypoint-based Monocular 3D Object Detection [article]

Lei Yang, Xinyu Zhang, Li Wang, Minghan Zhu, Jun Li
<span title="2021-06-12">2021</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
3D object detection with a single image is an essential and challenging task for autonomous driving.  ...  Recently, keypoint-based monocular 3D object detection has made tremendous progress and achieved great speed-accuracy trade-off.  ...  Another methods [3, 6, 15, 20, 26] infer 3D information using 2D/3D geometry constraint.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2105.00268v2">arXiv:2105.00268v2</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/5eovz5ygwngm3o2la3jpmgq6oa">fatcat:5eovz5ygwngm3o2la3jpmgq6oa</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20210616035753/https://arxiv.org/pdf/2105.00268v2.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/4e/c5/4ec56ecba51d531a369bda5e6b366d343a847e78.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2105.00268v2" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

M3D-RPN: Monocular 3D Region Proposal Network for Object Detection [article]

Garrick Brazil, Xiaoming Liu
<span title="2019-08-11">2019</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
Generally, the combination of expensive LiDAR sensors and stereo RGB imaging has been paramount for successful 3D object detection algorithms, whereas monocular image-only methods experience drastically  ...  M3D-RPN is able to significantly improve the performance of both monocular 3D Object Detection and Bird's Eye View tasks within the KITTI urban autonomous driving dataset, while efficiently using a shared  ...  We depict each parameter of within the 2D / 3D anchor formulation (left).  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/1907.06038v2">arXiv:1907.06038v2</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/pydrzhad6jg35ikzive5ze6qui">fatcat:pydrzhad6jg35ikzive5ze6qui</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20200906010751/https://arxiv.org/pdf/1907.06038v1.pdf" title="fulltext PDF download [not primary version]" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <span style="color: #f43e3e;">&#10033;</span> <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/82/4f/824f2da71b1348bcd3131ed61bdf0672c073ddc6.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/1907.06038v2" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

Kinematic 3D Object Detection in Monocular Video [article]

Garrick Brazil, Gerard Pons-Moll, Xiaoming Liu, Bernt Schiele
<span title="2020-07-19">2020</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
In this work, we propose a novel method for monocular video-based 3D object detection which carefully leverages kinematic motion to improve precision of 3D localization.  ...  Although temporal motion is an invaluable resource to human vision for detection, tracking, and depth perception, such features have not been thoroughly utilized in modern 3D object detectors.  ...  We first train the 2D-3D RPN with L = L 2D + L 3D , then the self-balancing loss of Eq. 7, for 80k and 50k iterations. We freeze the RPN to train ego-motion using L ego for 80k.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2007.09548v1">arXiv:2007.09548v1</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/yx4ky23qq5be3dva7ivcekghfm">fatcat:yx4ky23qq5be3dva7ivcekghfm</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20200812024827/https://arxiv.org/pdf/2007.09548v1.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2007.09548v1" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

CLOCs: Camera-LiDAR Object Candidates Fusion for 3D Object Detection [article]

Su Pang, Daniel Morris, Hayder Radha
<span title="2020-09-02">2020</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
There have been significant advances in neural networks for both 3D object detection using LiDAR and 2D object detection using video.  ...  Our experimental evaluation on the challenging KITTI object detection benchmark, including 3D and bird's eye view metrics, shows significant improvements, especially at long distance, over the state-of-the-art  ...  The four channel includes: IoU between 2D detections and projected 3D detections (IoU ), 2D confident score (s 2D ), 3D confident score (s 3D ) and normalized distance (d) between 3D bounding box and the  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2009.00784v1">arXiv:2009.00784v1</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/httyqfan4jet3hwa7uevi3q2uq">fatcat:httyqfan4jet3hwa7uevi3q2uq</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20200909003657/https://arxiv.org/pdf/2009.00784v1.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2009.00784v1" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

Monocular 3D Object Detection with Sequential Feature Association and Depth Hint Augmentation [article]

Tianze Gao, Huihui Pan, Huijun Gao
<span title="2020-12-02">2020</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
The 2D loss term is further adapted to be depth-aware for improving the detection accuracy of small objects.  ...  Monocular 3D object detection is a promising research topic for the intelligent perception systems of autonomous driving.  ...  The task of 3D object detection is then formulated as a PnP problem and solved by a 2D/3D matching algorithm [40] . (MonoGRNet) Qin et al.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2011.14589v3">arXiv:2011.14589v3</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/bjvmonr3j5dkppolcgdelmpjyy">fatcat:bjvmonr3j5dkppolcgdelmpjyy</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20201205220920/https://arxiv.org/pdf/2011.14589v3.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/dc/f3/dcf312e2f852d5abf6c6d20201e40238fb311fb3.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2011.14589v3" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

MonoJSG: Joint Semantic and Geometric Cost Volume for Monocular 3D Object Detection [article]

Qing Lian, Peiliang Li, Xiaozhi Chen
<span title="2022-03-16">2022</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
Due to the inherent ill-posed nature of 2D-3D projection, monocular 3D object detection lacks accurate depth recovery ability.  ...  Specifically, we first leverage neural networks to learn the object position, dimension, and dense normalized 3D object coordinates.  ...  To the best of our knowledge, we propose the first 2D-3D cost volume that computes the matching cost for object depth.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2203.08563v1">arXiv:2203.08563v1</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/szoxwdvvqbanxplb7jnc6vnmsy">fatcat:szoxwdvvqbanxplb7jnc6vnmsy</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20220321225744/https://arxiv.org/pdf/2203.08563v1.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/00/df/00df26281679ea7215e34ef3bee850e1348492b6.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2203.08563v1" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

FCOS3D: Fully Convolutional One-Stage Monocular 3D Object Detection [article]

Tai Wang, Xinge Zhu, Jiangmiao Pang, Dahua Lin
<span title="2021-09-24">2021</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
All of these make this framework simple yet effective, getting rid of any 2D detection or 2D-3D correspondence priors.  ...  Monocular 3D object detection is an important task for autonomous driving considering its advantage of low cost.  ...  Conclusion This paper proposes a simple yet efficient one-stage framework, FCOS3D, for monocular 3D object detection without any 2D detection or 2D-3D correspondence priors.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2104.10956v3">arXiv:2104.10956v3</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/ogyvb2gv4bhi5cyacu2giijrre">fatcat:ogyvb2gv4bhi5cyacu2giijrre</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20210929161701/https://arxiv.org/pdf/2104.10956v3.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/99/3e/993e389503ad2e4a473b5b0db3abe90f094c811d.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2104.10956v3" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

PillarNet: Real-Time and High-Performance Pillar-based 3D Object Detection [article]

Guangsheng Shi, Ruifeng Li, Chao Ma
<span title="2022-05-19">2022</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
Real-time and high-performance 3D object detection is of critical importance for autonomous driving.  ...  Additionally, PillarNet benefits from our designed orientation-decoupled IoU regression loss along with the IoU-aware prediction branch.  ...  Most existing methods discrete the sparse and irregular point clouds into regular grids including 3D voxels and 2D pillars, and then capitalize on 2D/3D CNN to perform object detection.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2205.07403v2">arXiv:2205.07403v2</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/7hpppd6r7jeajp3vfyqj4octuu">fatcat:7hpppd6r7jeajp3vfyqj4octuu</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20220524043909/https://arxiv.org/pdf/2205.07403v2.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/99/cd/99cddd2f738b80289323959b2c3c23a737702bb8.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2205.07403v2" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

Learning Depth-Guided Convolutions for Monocular 3D Object Detection [article]

Mingyu Ding, Yuqi Huo, Hongwei Yi, Zhe Wang, Jianping Shi, Zhiwu Lu, Ping Luo
<span title="2019-12-13">2019</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
Conventional 2D convolutions are unsuitable for this task because they fail to capture local object and its scale information, which are vital for 3D object detection.  ...  3D object detection from a single image without LiDAR is a challenging task due to the lack of accurate depth information.  ...  Finally, we outline the details of our 2D-3D detection head.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/1912.04799v2">arXiv:1912.04799v2</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/5gx5lr6o45dkhe3scq5jgb2ypa">fatcat:5gx5lr6o45dkhe3scq5jgb2ypa</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20200826152609/https://arxiv.org/pdf/1912.04799v2.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/88/41/8841036470c24823c2f33dcfbff855f12761876e.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/1912.04799v2" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>
&laquo; Previous Showing results 1 &mdash; 15 out of 581 results