GCN5 DG-STGCN: Dynamic Spatial-Temporal Modeling for Skeleton-based Action Recognition, ACM MM’22 논문 링크 : https://arxiv.org/pdf/2210.05895.pdf Abstract We note that existing GCN based approaches primarily rely on prescribed graphical structures (i.e., a manually defined topology of skeleton joints), which limits their flexibility to capture complicated correlations between joints. To move beyond this limitation, we propose a new framework for skeleton-based action recognition, namely Dynamic.. 2023. 7. 21. HOPE-Net: A Graph-based Model for Hand-Object Pose Estimation, CVPR’20 논문 링크 : https://openaccess.thecvf.com/content_CVPR_2020/papers/Doosti_HOPE-Net_A_Graph-Based_Model_for_Hand-Object_Pose_Estimation_CVPR_2020_paper.pdfAbstractHand-Object pose estimation(HOPE)는 손과 들고 있는 물체의 포즈를 동시에 감지하는 것을 목표로 한다. 본 논문에서는 2D와 3D에서 실시간으로 손과 물체의 포즈를 추정하는 HOPE-Net이라는 lightweight model을 제안한다. 여기에서는 두 가지의 adaptive graph convolution 을 사용하는데, 하나는 hand joint와 object corner의 2D 좌표를, 하나는 2.. 2023. 4. 6. Graph Attention Networks, ICLR’18 논문 링크 : https://arxiv.org/abs/1710.109031. IntroductionOn the other hand, we have non-spectral approaches (Duvenaud et al., 2015; Atwood & Towsley, 2016; Hamilton et al., 2017), which define convolutions directly on the graph, operating on groups of spatially close neighbors.One of the benefits of attention mechanisms is that they allow for dealing with variable sized inputs, focusing on the mos.. 2023. 3. 26. GaAN: Gated Attention Networks for Learning on Large and Spatiotemporal Graphs, UAI’18 논문 링크 : http://www.auai.org/uai2018/proceedings/papers/139.pdf1. IntroductionTreating each attention head equally loses the opportunity to benefit from some attention heads which are inherently more important than others. To this end, we propose the Gated Attention Networks(GaAN) for learning on graphs. GaAN uses a small convolutional subnetwork to compute a soft gate at each attention head to c.. 2023. 3. 26. 이전 1 2 다음