Pedestrian Trajectory Prediction Based on Improved Social Spatio-Temporal Graph Convolution Neural Network

Yang, Jikun; Han, Chao

doi:10.1145/3578741.3578754

Cited by 1 publication

(2 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Table 1 shows the ADE/FDE evaluation results of our proposed method on ETH and UCY datasets. Among them, models that utilize scene features included Sophie [28], DSCMP [29], RSBG [30], IST-PTEPN [31], and our method, while trajectory prediction methods based on graph convolution included STGAT [21], Social-STGCNN [19], AST-GNN [50], GCHGAT [51], PTP-STGCN [52], EvoSTGAT [53], Social TAG [42], Tri-HGNN [40], SKGACN [41], RDGCN [23], and our method. It can be concluded from Table 1 that our method outperforms all baseline methods in both average displacement error and final displacement error metrics.…”

Section: Quantitative Analysismentioning

confidence: 99%

“…The introduction of scene features is particularly important in achieving accurate prediction of pedestrian future trajectories. However, most of the previous trajectory prediction methods [28][29][30][31] that introduced scene features simply concatenated or added the scene features and trajectory representation features. Such methods cannot effectively model the potential relationship between scene features and pedestrian spatial-temporal interactions.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

WTGCN: Wavelet Transform Graph Convolution Network for Pedestrian Trajectory Prediction

Chen,

Sang,

Wang

et al. 2024

Preprint

View full text Add to dashboard Cite

The task of pedestrian trajectory prediction remains challenging due to variable scenarios, complex social interactions, and uncertainty in pedestrian motion. Previous trajectory prediction research only models from the time domain, which makes it difficult to accurately capture the global and detailed features of complex pedestrian social interactions and the uncertainty of pedestrian movement. These methods also ignore the relationship between scene features and the potential motion patterns of pedestrians. Therefore, we propose a wavelet transform graph convolution network to obtain accurate pedestrian potential motion patterns through time-frequency analysis. We first construct spatial and temporal graphs, then obtain the attention score matrices through the self-attention mechanism in the time domain and combine them with the scene features. Then, We utilize the two-dimensional discrete wavelet transform to generate low-frequency and high-frequency components for representing global and detailed features of spatial-temporal interactions. These components are then further processed using the asymmetric convolution, and the wavelet transform adjacency matrix is obtained through the inverse wavelet transform. We then employ graph convolution to combine the graph and the adjacency matrix to obtain spatial and temporal interaction features. Finally, we design the wavelet transform temporal convolution network to directly predict the two-dimensional Gaussian distribution parameters of the future trajectory. Extensive experiments on the ETH, UCY, and SDD datasets demonstrate that our method outperforms the state-of-the-art methods in prediction performance.

show abstract