Filters








7,890 Hits in 3.5 sec

ILabel: Interactive Neural Scene Labelling [article]

Shuaifeng Zhi and Edgar Sucar and Andre Mouton and Iain Haughton and Tristan Laidlow and Andrew J. Davison
2021 arXiv   pre-print
The scene model is updated and visualised in real-time, allowing the user to focus interactions to achieve efficient labelling.  ...  Joint representation of geometry, colour and semantics using a 3D neural field enables accurate dense labelling from ultra-sparse interactions as a user reconstructs a scene in real-time using a handheld  ...  The latter mode eases the burden of manual annotation, and users could provide labels via text or voice.  ... 
arXiv:2111.14637v2 fatcat:6rklf4bv5bam5gm2iaza5m557m

Semantic Instance Annotation of Street Scenes by 3D to 2D Label Transfer [article]

Jun Xie and Martin Kiefel and Ming-Ting Sun and Andreas Geiger
2016 arXiv   pre-print
A comparison of our method to state-of-the-art label transfer baselines reveals that 3D information enables more efficient annotation while at the same time resulting in improved accuracy and time-coherent  ...  Unfortunately, pixelwise annotation of images at very large scale is labor-intensive and only little labeled data is available, particularly at instance level and for street scenes.  ...  Method In this work, we are interested in generating semantic instance annotations for urban scenes at large scale by transferring labels from sparse 3D point clouds into the images.  ... 
arXiv:1511.03240v2 fatcat:ugcl4koyxzezrh3cvzfpvfuipu

Semantic Instance Annotation of Street Scenes by 3D to 2D Label Transfer

Jun Xie, Martin Kiefel, Ming-Ting Sun, Andreas Geiger
2016 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
A comparison of our method to state-ofthe-art label transfer baselines reveals that 3D information enables more efficient annotation while at the same time resulting in improved accuracy and time-coherent  ...  Unfortunately, pixelwise annotation of images at very large scale is labor-intensive and only little labeled data is available, particularly at instance level and for street scenes.  ...  Method In this work, we are interested in generating semantic instance annotations for urban scenes at large scale by transferring labels from sparse 3D point clouds into the images.  ... 
doi:10.1109/cvpr.2016.401 dblp:conf/cvpr/XieKSG16 fatcat:z52n32utwrgedchp3oe5hatybu

In-Place Scene Labelling and Understanding with Implicit Scene Representation [article]

Shuaifeng Zhi, Tristan Laidlow, Stefan Leutenegger, Andrew J. Davison
2021 arXiv   pre-print
We show the benefit of this approach when labels are either sparse or very noisy in room-scale scenes.  ...  The intrinsic multi-view consistency and smoothness of NeRF benefit semantics by enabling sparse labels to efficiently propagate.  ...  Another possible use case is in a scene labelling tool, since manual annotation in coarse images is much more efficient.  ... 
arXiv:2103.15875v2 fatcat:dkjq7aafl5f2rbunnkwxbtn64q

PointMatch: A Consistency Training Framework for Weakly Supervised Semantic Segmentation of 3D Point Clouds [article]

Yushuang Wu, Shengcai Cai, Zizheng Yan, Guanbin Li, Yizhou Yu, Xiaoguang Han, Shuguang Cui
2022 arXiv   pre-print
sparse points annotated.  ...  However, it suffers from (i) the inefficient exploitation of data information, and (ii) the strong reliance on labels thus is easily suppressed when given much fewer annotations.  ...  training signals to the whole scene via pseudo-labeling.  ... 
arXiv:2202.10705v2 fatcat:dxtcxnv3zjc7db6j7hrigf22ru

Box2Mask: Weakly Supervised 3D Semantic Instance Segmentation Using Bounding Boxes [article]

Julian Chibane, Francis Engelmann, Tuan Anh Tran, Gerard Pons-Moll
2022 arXiv   pre-print
The key idea is to leverage 3D bounding box labels which are easier and faster to annotate. Indeed, we show that it is possible to train dense segmentation models using only bounding box labels.  ...  This goes beyond commonly used center votes, which would not fully exploit the bounding box annotations.  ...  However, since scene and sub-scene labels are the coarsest annotations assumed, results typically lack details.  ... 
arXiv:2206.01203v2 fatcat:ea7r3w3i5nayhe22plh35mqpku

Automatic non-parametric image parsing via hierarchical semantic voting based on sparse–dense reconstruction and spatial–contextual cues

Xinyi An, Shuai Li, Hong Qin, Aimin Hao
2016 Neurocomputing  
leverage context-specific local-global label confidence propagation and global semantic spatial-contextual cues to guide holistic scene parsing.  ...  The originality of our new approach is to employ sparse-dense reconstruction as a latent learning model to conduct candidate-label probability analysis over multi-level local regions, and synchronously  ...  Then, we employ per-exemplar sparse-dense reconstruction for each region and determine the high-level region's label candidates via hierarchical voting.  ... 
doi:10.1016/j.neucom.2016.03.034 fatcat:vv6tz5t3wvdebgx5jpr4a6xm54

Author Index

2010 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition  
Veksler, Olga Tiered Scene Labeling with Dynamic Programming Vlachos  ...  Huang, Yuchi Image Retrieval via Probabilistic Hypergraph Ranking Automatic Image Annotation Using Group Sparsity Huang, Zhongyang Nonparametric Label-to-Region by Search Huber, Martin Lymph Node  ... 
doi:10.1109/cvpr.2010.5539913 fatcat:y6m5knstrzfyfin6jzusc42p54

Panoptic NeRF: 3D-to-2D Label Transfer for Panoptic Urban Scene Segmentation [article]

Xiao Fu, Shangzhan Zhang, Tianrun Chen, Yichong Lu, Lanyun Zhu, Xiaowei Zhou, Andreas Geiger, Yiyi Liao
2022 arXiv   pre-print
Unfortunately, pixel-wise annotation is labor-intensive and costly, raising the demand for more efficient labeling strategies.  ...  Experimental results show that Panoptic NeRF outperforms existing semantic and instance label transfer methods in terms of accuracy and multi-view consistency on challenging urban scenes of the KITTI-360  ...  We believe that our method is a step towards more efficient data annotation, while simultaneously providing a 3D consistent continuous panoptic representation of the scene.  ... 
arXiv:2203.15224v1 fatcat:oh6p55u4rzds5hblhs5sygiyva

KITTI-360: A Novel Dataset and Benchmarks for Urban Scene Understanding in 2D and 3D [article]

Yiyi Liao, Jun Xie, Andreas Geiger
2022 arXiv   pre-print
For efficient annotation, we created a tool to label 3D scenes with bounding primitives and developed a model that transfers this information into the 2D image domain, resulting in over 150k images and  ...  1B 3D points with coherent semantic instance annotations across 2D and 3D.  ...  Shrisha Bharadwaj, Apratim Bhattacharyya, Paul Henderson, and Zehao Yu for their help in implementing the baselines, Kashyap Chitta, Katja Schwarz, and Yue Wang for proofreading, and SurfingTech for annotating  ... 
arXiv:2109.13410v2 fatcat:dxqki3azibcobptngevkbz5ehq

One Thing One Click: A Self-Training Approach for Weakly Supervised 3D Semantic Segmentation [article]

Zhengzhe Liu, Xiaojuan Qi, Chi-Wing Fu
2021 arXiv   pre-print
To leverage these extremely sparse labels in network training, we design a novel self-training approach, in which we iteratively conduct the training and label propagation, facilitated by a graph propagation  ...  Point cloud semantic segmentation often requires largescale annotated training data, but clearly, point-wise labels are too tedious to prepare.  ...  Ours on "Data Efficient" 20 points/scene 59.4 Table 1.  ... 
arXiv:2104.02246v4 fatcat:ucniyqpo6zgi5k4ned7my5nafq

Training Constrained Deconvolutional Networks for Road Scene Semantic Segmentation [article]

German Ros, Simon Stent, Pablo F. Alcantarilla, Tomoki Watanabe
2016 arXiv   pre-print
To address the first constraint, we introduce a Multi-Domain Road Scene Semantic Segmentation (MDRS3) dataset, aggregating data from six existing densely and sparsely labelled datasets for training our  ...  In this work we investigate the problem of road scene semantic segmentation using Deconvolutional Networks (DNs).  ...  Refinement of sparse annotations.  ... 
arXiv:1604.01545v1 fatcat:flm7e6sd35av7c45ycvr7zx5nu

Language-Grounded Indoor 3D Semantic Segmentation in the Wild [article]

David Rozenberszki, Or Litany, Angela Dai
2022 arXiv   pre-print
consistently outperforms state-of-the-art 3D pre-training for 3D semantic segmentation on our proposed benchmark (+9% relative mIoU), including limited-data scenarios with +25% relative mIoU using only 5% annotations  ...  In addition considering the setting where all dense annotations are available for train scenes for the 200 classes, we also consider limited annotation scenarios with only sparse annotations per scene,  ...  Hou, J., Graham, B., Nießner, M., Xie, S.: Exploring data-efficient 3d scene understanding with contrastive scene contexts.  ... 
arXiv:2204.07761v2 fatcat:axlrozwpnjejnaarmeqrxyvgla

Guest Editorial: Big Data

Alyosha Efros, Antonio Torralba
2016 International Journal of Computer Vision  
"Sparse Output Coding for Scalable Visual Recognition" deals with situations when the big data also has high cardinality in the number of classes, proposing an efficient sparse-coding scheme.  ...  label propagation.  ... 
doi:10.1007/s11263-016-0914-5 fatcat:76kfkzj6c5dste6ct2kzrg4sta

A Convex Optimization Framework for Active Learning

Ehsan Elhamifar, Guillermo Sapiro, Allen Yang, S. Shankar Sasrty
2013 2013 IEEE International Conference on Computer Vision  
In this paper, we develop an efficient active learning framework based on convex programming, which can select multiple samples at a time for annotation.  ...  However, it is typically expensive and time consuming to obtain labels for the samples.  ...  Active Learning via Convex Programming In this section, we propose an efficient algorithm for active learning that takes advantage of convex programming in order to find the most informative points.  ... 
doi:10.1109/iccv.2013.33 dblp:conf/iccv/ElhamifarSYS13 fatcat:fbba3s3s3fcnxckxsxrxlooh7m
« Previous Showing results 1 — 15 out of 7,890 results