A Thorough Benchmark and a New Model for Light Field Saliency Detection

Gao, Wei; Fan, Songlin; Li, Ge; Wang, Lin

doi:10.1109/tpami.2023.3235415

Cited by 25 publications

(14 citation statements)

References 73 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this paper, we propose a learning-based method using data augmentation for light field SOD. We take the recently proposed PKU-LF 7 for data augmentation and use it as part of our training set. Then a mutual attention method is used to extract and fuse features from RGB images and depth maps.…”

Section: Introductionmentioning

confidence: 99%

A learning-based method using data augmentation for light field salient object detection

Zhu,

Wang,

Zheng

2023

Optoelectronic Imaging and Multimedia Technology X

View full text Add to dashboard Cite

Section: Introductionmentioning

confidence: 99%

A learning-based method using data augmentation for light field salient object detection

Zhu,

Wang,

Zheng

2023

Optoelectronic Imaging and Multimedia Technology X

View full text Add to dashboard Cite

“…This asynchronous event-based feature offers several advantages, including high temporal resolution, low power consumption, and the potential for privacy encryption. While several methods have explored action recognition using event data in single-view settings [13], [14], [15], [16], there is currently no existing work that utilizes multi-view event data for action recognition to the best of our knowledge.…”

Section: Introductionmentioning

confidence: 99%

“…One of the reasons is the lack of datasets. Although there are existing single-view event action datasets [16], [17], [18], [19], there is a lack of comprehensive multiview event datasets specifically designed for action recognition. DHP19 [18] is the only dataset that can be used for multi-view event-based action recognition, but it is oriented towards pose estimation tasks and is small in scale (33 actions and 2,228 recordings).…”

Section: Introductionmentioning

confidence: 99%

“…DHP19 [18] is the only dataset that can be used for multi-view event-based action recognition, but it is oriented towards pose estimation tasks and is small in scale (33 actions and 2,228 recordings). To facilitate research in multi-view event-based action recognition, we introduce the THU MV-EACT -50, an expansion of the singleview THU E-ACT -50 [16], by incorporating more viewpoints. The THU MV-EACT -50 dataset comprises 50 distinct actions observed from 6 different views, encompassing 4 frontal and 2 backal views, resulting in a comprehensive collection of 31,500 recording sequences.…”

Section: Introductionmentioning

confidence: 99%

“…• We extend single-view THU E-ACT -50 [16] to contribute the THU MV-EACT -50 dataset, the largest multiview event action dataset to date, comprising 50 actions from 6 viewpoints, providing a valuable resource for evaluating algorithms in multiview event-based action recognition. The constructed benchmarks can be accessed at: https://gaoyue.org/ dataset/THU-MV-EACT-50.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Hypergraph Neural Network for Skeleton-Based Action Recognition

Hao

Guo

et al. 2021

IEEE Trans. on Image Process.

View full text Add to dashboard Cite

Action recognition from video data forms a cornerstone with wide-ranging applications. Single-view action recognition faces limitations due to its reliance on a single viewpoint. In contrast, multi-view approaches capture complementary information from various viewpoints for improved accuracy. Recently, event cameras have emerged as innovative bio-inspired sensors, leading to advancements in event-based action recognition. However, existing works predominantly focus on single-view scenarios, leaving a gap in multi-view event data exploitation, particularly in challenges like information deficit and semantic misalignment. To bridge this gap, we introduce HyperMV, a multi-view event-based action recognition framework. HyperMV converts discrete event data into frame-like representations and extracts view-related features using a shared convolutional network. By treating segments as vertices and constructing hyperedges using rule-based and KNN-based strategies, a multi-view hypergraph neural network that captures relationships across viewpoint and temporal features is established. The vertex attention hypergraph propagation is also introduced for enhanced feature fusion. To prompt research in this area, we present the largest multi-view event-based action dataset THU MV-EACT -50, comprising 50 actions from 6 viewpoints, which surpasses existing datasets by over tenfold. Experimental results show that HyperMV significantly outperforms baselines in both cross-subject and cross-view scenarios, and also exceeds the state-of-the-arts in frame-based multi-view action recognition.

show abstract