Otwarty dostęp

Artistic Expressions and Aesthetic Characteristics of Combining Dance and Folk Music

,  oraz   
25 wrz 2025

Zacytuj
Pobierz okładkę

Figure 1.

DeepLabv3+ model structure
DeepLabv3+ model structure

Figure 2.

Decoding layer network of DeepLabV3+
Decoding layer network of DeepLabV3+

Figure 3.

LSTM dance activity recognition network model structure
LSTM dance activity recognition network model structure

Figure 4.

Results of action segmentation
Results of action segmentation

Figure 5.

Results of user satisfaction
Results of user satisfaction

Results of video segmentation of dance movements

Fragment number Action description Start frame/number End frame/number
1 Bend the knee 1 183
2 High leg lift 184 251
3 Twist the waist 252 304
4 Rotation 305 387
5 Step on toes 388 413
6 Wave your hand 414 498

Experimental results on the NTU-RGB+D60 dataset(%)

Model Cross-Subject Cross-View
HBRNN 54.5 61.9
ST-GCN 66.4 79.5
RA-GCNv2 86.3 90.1
AS-GCN 88.1 91.5
2s-AGCN 89.7 93.1
DGNN 89.9 93.8
PL-GCN 90.2 94.6
DC-GCN-ADC 90.9 95.0
PA-ResGCN-B19 91.3 95.2
Dynamic-GCN 91.5 96.3
MST-GCN 90.2 95.9
Hyper-GNN 91.4 96.5
STF-HGCN 88.7 96.0
The proposed 94.3 97.9

Comparison of dance movement assessment

Model PFC BAS Distk Distg
The proposed 1.357 0.178 10.699 7.663
DTW 1.736 0.152 11.397 7.513
Euclidean-DTW 1.578 0.135 10.995 7.536
BLAST-DTW 1.682 0.141 11.098 7.528
True value 1.275 0.182 10.738 7.692

Experimental results on the NTU-RGB+D120 dataset(%)

Model Cross-Subject Cross-setup
ST-GCN 66.3 69.1
RA-GCNv2 79.5 80.1
2s-AGCN 81.7 84.3
SGN 76.5 72.4
Shift-GCN 84.8 86.9
MS-G3D 87.3 88.2
DC-GCN+ADC 75.6 76.1
PA-ResGCN-B19 80.4 83.2
Dynamic-GCN 86.1 87.5
STF-HGCN 85.4 88.2
The proposed 89.3 90.2
Język:
Angielski
Częstotliwość wydawania:
1 razy w roku
Dziedziny czasopisma:
Nauki biologiczne, Nauki biologiczne, inne, Matematyka, Matematyka stosowana, Matematyka ogólna, Fizyka, Fizyka, inne