{"slug": "trajfusionnet-transformer-based-prediction-of-pedestrian-crossing-intention-via", "title": "TrajFusionNet+: Transformer-Based Prediction of Pedestrian Crossing Intention via Fusion of Trajectory Representations and Scene Graphs", "summary": "Researchers introduced TrajFusionNet+, a transformer-based model that predicts pedestrian crossing intention for autonomous vehicles by fusing sequential trajectory data, visual bounding-box overlays, and pedestrian-centric scene graphs. TrajFusionNet+ achieved improved state-of-the-art performance on the PIE and JAAD datasets, and under a new protocol training jointly on both datasets while evaluating separately, the model showed superior generalization to existing approaches. The architecture comprises a Sequence Attention Module, a Visual Attention Module, and a Graph Attention Module, building on the team's earlier TrajFusionNet model.", "body_md": "arXiv:2609.10806v1 Announce Type: new \nAbstract: The pedestrian crossing intention task involves predicting whether pedestrians are likely to cross the road from the point of view of an autonomous vehicle. We introduce TrajFusionNet+, a novel transformer-based model for pedestrian crossing intention prediction. TrajFusionNet+ combines sequential and visual representations of pedestrian trajectory with a graph-based representation of the scene context in order to predict pedestrian crossing intention. The proposed architecture builds upon our previous model, TrajFusionNet, and comprises three branches: a Sequence Attention Module (SAM), which processes a sequential representation of past and predicted pedestrian trajectories; a Visual Attention Module (VAM), which utilizes a visual representation of the pedestrian trajectories by overlaying observed and predicted bounding boxes onto scene images; and a Graph Attention Module (GAM), which extracts pedestrian-centric graphs from segmented scene images and captures the relational dependencies between pedestrians and traffic elements. TrajFusionNet+ achieves improved state-of-the-art performance on the two most widely used pedestrian crossing intention datasets, PIE and JAAD. Furthermore, we introduce a new evaluation protocol in which models are trained jointly on the PIE and JAAD datasets but evaluated separately on each. Under this setting, TrajFusionNet+ demonstrates superior generalization compared to existing approaches.", "url": "https://wpnews.pro/news/trajfusionnet-transformer-based-prediction-of-pedestrian-crossing-intention-via", "canonical_source": "https://arxiv.org/abs/2609.10806", "published_at": "2026-09-11 04:00:00+00:00", "updated_at": "2026-09-11 04:29:21.301111+00:00", "lang": "en", "topics": ["autonomous-vehicles", "computer-vision", "artificial-intelligence", "machine-learning", "ai-research"], "entities": ["TrajFusionNet+", "TrajFusionNet", "Sequence Attention Module", "Visual Attention Module", "Graph Attention Module", "PIE dataset", "JAAD dataset"], "alternates": {"html": "https://wpnews.pro/news/trajfusionnet-transformer-based-prediction-of-pedestrian-crossing-intention-via", "markdown": "https://wpnews.pro/news/trajfusionnet-transformer-based-prediction-of-pedestrian-crossing-intention-via.md", "text": "https://wpnews.pro/news/trajfusionnet-transformer-based-prediction-of-pedestrian-crossing-intention-via.txt", "jsonld": "https://wpnews.pro/news/trajfusionnet-transformer-based-prediction-of-pedestrian-crossing-intention-via.jsonld"}}