NMS Strikes Back

Ouyang-Zhang, Jeffrey; Cho, Jang Hyun

doi:10.48550/arxiv.2212.06137

Cited by 4 publications

(5 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Nevertheless, the above one-toone assignment in vanilla Hungarian matching algorithm would cause the severe separator missing problem, due to the slim shape of separator. To attack it, inspired by method [30], we further modify the assignment to the oneto-many strategy, i.e., one GT is assigned with a group rather one separator. Based on the one-to-one assigned separator {s γi , c γi }, we further find its neighboring ones on the condition that D sep < 5.…”

Section: Training Strategymentioning

confidence: 99%

Neural Collaborative Graph Machines for Table Structure Recognition

Líu¹,

Li²,

Liu³

et al. 2022

2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

View full text Add to dashboard Cite

Recently, Table Structure Recognition (TSR) task, aiming at identifying table structure into machine readable formats, has received increasing interest in the community. While impressive success, most single table componentbased methods can not perform well on unregularized table cases distracted by not only complicated inner structure but also exterior capture distortion. In this paper, we raise it as Complex TSR problem, where the performance degeneration of existing methods is attributable to their inefficient component usage and redundant post-processing. To mitigate it, we shift our perspective from table component extraction towards the efficient multiple components leverage, which awaits further exploration in the field. Specifically, we propose a seminal method, termed GrabTab, equipped with newly proposed Component Deliberator. Thanks to its progressive deliberation mechanism, our GrabTab can flexibly accommodate to most complex tables with reasonable components selected but without complicated postprocessing involved. Quantitative experimental results on public benchmarks demonstrate that our method significantly outperforms the state-of-the-arts, especially under more challenging scenes.

show abstract

Section: Training Strategymentioning

confidence: 99%

Neural Collaborative Graph Machines for Table Structure Recognition

Líu¹,

Li²,

Liu³

et al. 2022

2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

View full text Add to dashboard Cite

show abstract

“…DsLMF+ dataset have provided two types of datasets formats of YOLO and COCO, which make it convenient to be applied for the currently popular top-ranked target detection neural networks. In order to verify the feasibility of the constructed dataset, this work selected YOLOv7 34 , DETA 35 and ViT-Adapter-L 36 www.nature.com/scientificdata www.nature.com/scientificdata/ on the DsLMF+ dataset. The access links of DETA, ViT-Adapter-L and YOLOv7 that used to verify the datasets are respectively https://github.com/jozhang97/deta, https://github.com/czczup/vit-adapter and https://github.…”

Section: Technical Validationmentioning

confidence: 99%

An open dataset for intelligent recognition and classification of abnormal condition in longwall mining

Yang

Zhang

et al. 2023

Sci Data

View full text Add to dashboard Cite

The underground coal mine production of the fully mechanized mining face exists many problems, such as poor operating environment, high accident rate and so on. Recently, the intelligent autonomous coal mining is gradually replacing the traditional mining process. The artificial intelligence technology is an active research area and is expect to identify and warn the underground abnormal conditions for intelligent longwall mining. It is inseparable from the construction of datasets, but the downhole dataset is still blank at present. This work develops an image dataset of underground longwall mining face (DsLMF+), which consists of 138004 images with annotation 6 categories of mine personnel, hydraulic support guard plate, large coal, towline, miners’ behaviour and mine safety helmet. All the labels of dataset are publicly available in YOLO format and COCO format. The availability and accuracy of the datasets were reviewed by experts in coal mine field. The dataset is open access and aims to support further research and advancement of the intelligent identification and classification of abnormal conditions for underground mining.

show abstract

“…In [74], the authors argue that the Hungarian loss's direct one-to-one bounding box matching approach might not always be advantageous. They demonstrate that employing a one-tomany assignment strategy and utilizing the NMS (Non-Maximum Suppression) module leads to better detection results.…”

Section: Architecture and Block Modificationsmentioning

confidence: 99%

Extended Expectation Maximization for Under-Fitted Models

Rekavandi¹,

SEGHOUANE²,

Boussaid³

et al. 2022

Preprint

View full text Add to dashboard Cite

<p>In this paper, we generalize the well-known Expectation Maximization (EM) algorithm using the α− divergence for Gaussian Mixture Model (GMM). This approach is used in robust subspace detection when the number of parameters is kept small to avoid overfitting and large estimation variances. The level of robustness can be tuned by the parameterα. When α→1, our method is equivalent to the standard EM approach and for α <1 the method is robust against potential outliers. Simulation results show that the method outperforms the standard EM when it comes to mismatches between noise models and their realizations. In addition, we use the proposed method to detect active brain areas using collected functional Magnetic Resonance Imaging (fMRI) data during task-related experiments.</p>

show abstract

NMS Strikes Back

Cited by 4 publications

References 0 publications

Neural Collaborative Graph Machines for Table Structure Recognition

Neural Collaborative Graph Machines for Table Structure Recognition

An open dataset for intelligent recognition and classification of abnormal condition in longwall mining

Extended Expectation Maximization for Under-Fitted Models

Contact Info

Product

Resources

About