Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces HDST-GNN, a novel Heterogeneous Dynamic Spatiotemporal Graph Neural Network designed to tackle the complexities of multi-object tracking (MOT) in UAV aerial imagery. By incorporating altitude-adaptive edge construction, heterogeneous node representations, and occlusion-gated temporal aggregation, the model effectively addresses challenges such as varying altitudes, object occlusions, and identity switches. The results demonstrate a significant performance improvement over existing methods, achieving 94.51% MOTA and reducing identity switches by 81% compared to SORT, highlighting the effectiveness of its tailored approach to MOT.
HDST-GNN outperforms traditional methods by reducing identity switches by 81% through innovative handling of occlusions and heterogeneous object states.
Multi-object tracking (MOT) from UAV imagery presents unique challenges: altitude varies across sequences, objects are small and densely packed, and frequent occlusion causes identity switches. Existing graph-based trackers assume fixed spatial context and treat all objects uniformly, ignoring the heterogeneous lifecycle states of detections, active tracklets, and lost targets. We propose HDST-GNN, a Heterogeneous Dynamic Spatiotemporal Graph Neural Network with three novel contributions. First, Altitude-Adaptive Edge Construction estimates a camera-altitude proxy from mean object area and adjusts the graph connectivity radius accordingly. Second, Heterogeneous Node Representation models detections (Type-D), confirmed tracklets (Type-T), and lost tracklets (Type-L) as distinct node types with dedicated projections and typed edge relations. Third, Occlusion-Gated Temporal Aggregation gates each node's attention contribution by its occlusion confidence, preventing occluded nodes from corrupting neighbour embeddings. HDST-GNN is trained end-to-end with a differentiable Sinkhorn head using joint cross-entropy and triplet loss. On VisDrone2019-MOT with oracle detections, HDST-GNN achieves 94.51% MOTA and 97.24% IDF1, outperforming SORT by +5.0 MOTA points and reducing identity switches by 81%. With real YOLOv8n detections, HDST-GNN reduces identity switches by 49% vs. SORT. Ablation studies confirm the independent contribution of each component.