CAD: concatenated action descriptor for one and two person(s), using silhouette and silhouette's skeleton

Islam, M. Shujah; Iqbal, Mansoor; Naqvi, Nuzhat; Bakhat, Khush; Islam, Muhammad; Kanwal, Shamsa; Ye, Zhongfu

doi:10.1049/iet-ipr.2018.6437

Cited by 10 publications

(4 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Skeletal data [34] Deep LSTM [35] Hybrid feature [36] Hybrid descriptors [37] OMKZ [38] 88 90.41 91.25 91.63 93.80 CAD [39] 95.6 FGP-3D 99.5…”

Section: Methods Accuracy (%)mentioning

confidence: 99%

“…Performance is described as classification accuracy on the FGP-3D of the SBU dataset using only the combined data unless otherwise stated [43]. As shown in Table 4, we compare our model to state-of-the-art models [34][35][36][37][38][39] on the dataset before analyzing how much performance increases on target datasets. On all datasets, the final proposed FGP-3D models are assessed in comparison to other cutting-edge methods for action recognition.…”

Section: Methods Accuracy (%)mentioning

confidence: 99%

“…Instead of taking advantage of fine-tuning action recognition models, previous techniques were only tested on target datasets. Recent research [31][32][33]38,39] proposed view-invariant using 2D or 3D incorporating algorithms with joints performed on our evaluated datasets [25,[31][32][33][34][35]38,39] that do not correspond to legit, and thus these techniques struggle to improve the action recognition performance on regression tasks with practical and mostly used databases [40][41][42][43]. These algorithms were developed to investigate the transferability of action recognition using the human skeleton.…”

Section: Related Workmentioning

confidence: 99%

“…In Table 2, cuttingedge methods [25][26][27][28][29] and state-of-the-art approaches [28,[30][31][32][33] are contrasted in Table 3. The state-of-the-art approaches [34][35][36][37][38][39] are contrasted in Table 4. Our technique surpasses all other methods in every evaluation scenario on all four challenge datasets.…”

Section: Evaluation Of Fgp-3dmentioning

confidence: 99%

See 3 more Smart Citations

Employing FGP-3D, a Fully Gated and Anchored Methodology, to Identify Skeleton-Based Action Recognition

et al. 2023

Self Cite

View full text Add to dashboard Cite

Recent years have seen an explosion in interest in and development of action recognition based on skeletal data. Contemporary methods using fully gated units can successfully extract characteristics from human skeletons by relying on the human topology that has been predefined. Despite advancements, fully gated unit-based techniques have trouble generalizing to other domains, particularly when dealing with various human topological structures. In this context, we introduce FGP-3D, a novel skeleton-based action recognition technique that can generalize across datasets while being effective at learning spatiotemporal features from human skeleton sequences. This is accomplished via a multi-head attention technique to learn an ideal dependence feature matrix from the uniform distribution. We next re-evaluate state-of-the-art techniques as well as the suggested novel descriptor FGP-3D in order to examine the cross-domain generalizability of skeleton-based action recognition in real-world video skeleton statistics. After being applied to commonly used action categorization datasets, experimental results demonstrate that the proposed FGP-3D, with pre-training, generalizes well and outperforms the state-of-the-art.

show abstract

“…Skeletal data [34] Deep LSTM [35] Hybrid feature [36] Hybrid descriptors [37] OMKZ [38] 88 90.41 91.25 91.63 93.80 CAD [39] 95.6 FGP-3D 99.5…”

Section: Methods Accuracy (%)mentioning

confidence: 99%