Trajectory-Aware Adaptive Inference in Object Detection Models

2026-05-12

AI Summary

This work addresses the challenge of achieving both high accuracy and computational efficiency in real-time object detection within dynamic maritime environments, where existing methods struggle to effectively balance trajectory awareness with inference cost. To this end, the authors propose a trajectory-aware adaptive inference mechanism that, for the first time, integrates motion cues—such as inter-vessel distances and relative velocities—into the YOLOv8 detector. This integration enables dynamic assessment of scene complexity and triggers an early-exit strategy that selectively activates either a subnetwork or the full model based on current conditions. The proposed approach significantly reduces inference latency and computational overhead while preserving high detection accuracy, thereby enabling a flexible trade-off between precision and efficiency.

Abstract

The increasing integration of sensors in autonomous maritime navigation has led to large-scale multimodal datasets, raising challenges in achieving efficient real-time perception. In such systems, object detection and trajectory perception of nearby vessels are tightly coupled, particularly in dynamic environments such as maritime navigation. However, the efficiency of object detection models during inference remains an often-overlooked aspect. To this end, we build upon an existing object detection framework by incorporating GPS trajectory data into the inference process to enable input-adaptive computation. Specifically, we introduce an early-exit mechanism in a YOLOv8-based detector that incorporates motion cues - such as inter-vessel distances. Frames of vessels that are separated by short distances, converging with high speed, are processed using the full model, while only a subset of the network's architecture is activated otherwise. The difficulty degree (or scene complexity) of a frame or set of frames per second is evaluated by leveraging inter-object distance and the rate at which the distance between them decreases. Experimental results demonstrate that this strategy maintains satisfactory detection performance while significantly reducing inference time and computational cost, thus enabling a flexible trade-off between accuracy and efficiency compared to full-model inference.

Problem

• object detection
• inference efficiency
• autonomous maritime navigation
• real-time perception
• trajectory-aware

Innovation

• trajectory-aware inference
• adaptive computation
• early-exit mechanism
• YOLOv8
• scene complexity