119,754 Hits in 7.3 sec

Semantic Segmentation from Limited Training Data [article]

A. Milan, T. Pham, K. Vijay, D. Morrison, A.W. Tow, L. Liu, J. Erskine, R. Grinover, A. Gurman, T. Hunn, N. Kelly-Boxall, D. Lee (+13 others)
2017 arXiv   pre-print
In contrast to traditional approaches which require large collections of annotated data and many hours of training, the task here was to obtain a robust perception pipeline with only few minutes of data  ...  The other is a fully-supervised semantic segmentation approach with efficient dataset collection. We conduct an extensive analysis of the two methods on our ARC 2017 dataset.  ...  While this is fairly straightforward in cases where all categories are known and defined beforehand, it is not such a common strategy for tasks where the amount of available training data is very limited  ... 
arXiv:1709.07665v1 fatcat:syfujvprnvbjrhob5tvy2syrpq

Semantic Segmentation from Limited Training Data

A. Milan, T. Pham, K. Vijay, D. Morrison, A.W. Tow, L. Liu, J. Erskine, R. Grinover, A. Gurman, T. Hunn, N. Kelly-Boxall, D. Lee (+13 others)
2018 2018 IEEE International Conference on Robotics and Automation (ICRA)  
While this is fairly straightforward in cases where all categories are known and defined beforehand, it is not such a common strategy for tasks where the amount of available training data is very limited  ...  We capture seven images of each item from the competition set before each official run to finetune our semantic segmentation approach.  ... 
doi:10.1109/icra.2018.8461082 dblp:conf/icra/MilanPVMTLEGGHK18 fatcat:773po3dfmfek7lvulqacevinfy

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

David Rozenberszki, Or Litany, Angela Dai
2022 arXiv   pre-print
Extensive experiments show that our approach consistently outperforms state-of-the-art 3D pre-training for 3D semantic segmentation on our proposed benchmark (+9% relative mIoU), including limited-data  ...  To learn more robust 3D features in this context, we propose a language-driven pre-training method to encourage learned 3D features that might have limited training examples to lie close to their pre-trained  ...  Experiments on our ScanNet200 semantic segmentation as well as semantic segmentation in the limited data regime demonstrate the effectiveness of our language-grounded 3D semantic segmentation.  ... 
arXiv:2204.07761v2 fatcat:axlrozwpnjejnaarmeqrxyvgla

Satellite Image Semantic Segmentation [article]

Eric Guérin, Killian Oechslin, Christian Wolf, Benoît Martinez
2021 arXiv   pre-print
We rely on Swin Transformer architecture and build the dataset from IGN open data. We report quantitative and qualitative segmentation results on this dataset and discuss strengths and limitations.  ...  In this paper, we propose a method for the automatic semantic segmentation of satellite images into six classes (sparse forest, dense forest, moor, herbaceous formation, building, and road).  ...  Limitations As illustrated in the previous images, the results are not perfect. This is caused by the inherent limits of the data used during the training phase. The main limitations are: 1.  ... 
arXiv:2110.05812v1 fatcat:sx6rxjse7fenlhxfmv2adxrsyq


Y. Cao, M. Scaioni
2021 The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences  
By supplying a small amount of task-specific supervision, a segmentation network is proposed for semantically segmenting the encoded features acquired from the pre-trained AE.  ...  Consequently, the application of fully supervised DL for semantic segmentation of buildings' point clouds at LoD3 level is severely limited.  ...  ACKNOWLEDGEMENTS Financial support from the program of China Scholarships Council (Grant No. 201906860014) is acknowledged. We thank Dr.Matrone et al. for the ArCH dataset.  ... 
doi:10.5194/isprs-archives-xliii-b2-2021-449-2021 fatcat:36l26igkjjaoznwf3kqc3ibmhm

Exploring Data-Efficient 3D Scene Understanding with Contrastive Scene Contexts [article]

Ji Hou, Benjamin Graham, Matthias Nießner, Saining Xie
2021 arXiv   pre-print
Our method achieves state-of-the-art results on a suite of benchmarks where training data or labels are scarce.  ...  Our study reveals that exhaustive labelling of 3D point clouds might be unnecessary; and remarkably, on ScanNet, even using 0.1% of point labels, we still achieve 89% (instance segmentation) and 96% (semantic  ...  Compared to 2D vision, the limits of big data are far from being fully explored in 3D. In 2D representation learning, for example, transfer learning from a rich source data (e.g.  ... 
arXiv:2012.09165v3 fatcat:oadjpimlvfgvljr737yxh2oijy

Building medical image classifiers with very limited data using segmentation networks

Ken C.L. Wong, Tanveer Syeda-Mahmood, Mehdi Moradi
2018 Medical Image Analysis  
By using a segmentation network pre-trained on similar data as the classification task, the machine can first learn the simpler shape and structural concepts before tackling the actual classification problem  ...  Comparisons with ImageNet pre-trained classifiers and classifiers trained from scratch are presented.  ...  used with limited data.  ... 
doi:10.1016/ pmid:30119038 fatcat:uox5strguvdd3ncuoz7ecmihu4

Learning High-Resolution Domain-Specific Representations with a GAN Generator [article]

Danil Galeev, Konstantin Sofiiuk, Danila Rukhovich, Mikhail Romanov, Olga Barinova, Anton Konushin
2020 arXiv   pre-print
Moreover, this simple approach also outperforms recent semi-supervised semantic segmentation methods that use both labeled and unlabeled data during training.  ...  We find that such semantic projection can be learnt from just a few annotated images.  ...  The semantic segmentation model is trained using available annotated data and does not use the unlabeled data.  ... 
arXiv:2006.10451v1 fatcat:bzbzgpcqhfgexbwkudnpnistie

Digging into Pseudo Label: a Low-budget Approach for Semi-Supervised Semantic Segmentation

Zhenghao Chen, Rui Zhang, Gang Zhang, Zhenhuan Ma, Tao Lei
2020 IEEE Access  
Although semi-supervised learning for image classification has been extensively studied in some cases, semantic segmentation with limited data has only recently gained attention.  ...  reliable supervision information from pseudo-labels to assist in training network with strong labels.  ...  [22] proposed a GAN-based segmentation network that can learn beneficial feature information from unlabeled data to assist in the training of labeled data. On the other hand, Bellver et al.  ... 
doi:10.1109/access.2020.2975022 fatcat:3gicsfklwbc7deurtryl655ovy

Universal Semi-Supervised Semantic Segmentation [article]

Tarun Kalluri, Girish Varma, Manmohan Chandraker, C V Jawahar
2019 arXiv   pre-print
In recent years, the need for semantic segmentation has arisen across several different applications and environments.  ...  However, the expense and redundancy of annotation often limits the quantity of labels available for training in any domain, while deployment is easier if a single model works well across domains.  ...  limited training data remains.  ... 
arXiv:1811.10323v3 fatcat:gzblvf2f4ndndamvpoxda57asu

Learning Pixel-wise Labeling from the Internet without Human Interaction [article]

Yun Liu, Yujun Shi, JiaWang Bian, Le Zhang, Ming-Ming Cheng, Jiashi Feng
2018 arXiv   pre-print
query keywords for segmentation model training.  ...  Collecting sufficient annotated data is very expensive in many applications, especially for pixel-level prediction tasks such as semantic segmentation.  ...  Compared with previous weakly-supervised semantic segmentation [13, 25, 27, 9, 28, [35] [36] [37] 12] that is limited to pre-defined categories due to the limitation of human-annotated training data,  ... 
arXiv:1805.07548v1 fatcat:lvq5c2ts2zhenafjguuatxc3uu

Data Efficient 3D Learner via Knowledge Transferred from 2D Model [article]

Ping-Chung Yu, Cheng Sun, Min Sun
2022 arXiv   pre-print
On ScanNet official evaluation, we establish new state-of-the-art semantic segmentation results on the data-efficient track.  ...  Specifically, we utilize a strong and well-trained semantic segmentation model for 2D images to augment RGB-D images with pseudo-label. The augmented dataset can then be used to pre-train 3D models.  ...  Shape analysis under limited data scenario Our approach discussed above presents a positive outcome on scene-level semantic segmentation under the limited data scenario.  ... 
arXiv:2203.08479v2 fatcat:4xhrrwld7ngs3kz4b6ry6ba364

InstantDL: an easy-to-use deep learning pipeline for image segmentation and classification

Dominik Jens Elias Waibel, Sayedali Shetab Boushehri, Carsten Marr
2021 BMC Bioinformatics  
We have thus developed InstantDL, a deep learning pipeline for four common image processing tasks: semantic segmentation, instance segmentation, pixel-wise regression and classification.  ...  Convolutional neural networks have become the state-of-the-art tool to provide accurate and fast image data processing.  ...  the nuclei in microscopy images of multiple organs segmentation, respectively [14, 15]), two pre-trained weights from 2D lung segmentation [16] (from semantic and instance segmentation), two from 3D  ... 
doi:10.1186/s12859-021-04037-3 pmid:33653266 fatcat:olqnsuxlxjgwjmrc5dqdcvva4u

Semantic Implicit Neural Scene Representations With Semi-Supervised Training [article]

Amit Kohli, Vincent Sitzmann, Gordon Wetzstein
2021 arXiv   pre-print
Our method is simple, general, and only requires a few tens of labeled 2D segmentation masks in order to achieve dense 3D semantic segmentation.  ...  We take the next step and demonstrate that an existing implicit representation (SRNs) is actually multi-modal; it can be further leveraged to perform per-point semantic segmentation while retaining its  ...  In this case of limited training data, we parameterize SEG as a linear classifier in order to prevent overfitting.  ... 
arXiv:2003.12673v2 fatcat:pcijezws6vhznmz4ih5buuytva


S. Ham, Y. Oh, K. Choi, I. Lee
2018 The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences  
Specifically, we train a deconvolutional network with publicly opened geospatial data, semantically segment a given UAV image into a building probability map and compare the building map with existing  ...  GIS data.  ...  We also found that a deconvolutional network trained with the national geospatial data can be used for semantic segmentation of UAV images.  ... 
doi:10.5194/isprs-archives-xlii-2-419-2018 fatcat:kykictfzkrf5bfu7gtccv4sc7e
« Previous Showing results 1 — 15 out of 119,754 results