YOLOv8-SMOT: An Efficient and Robust Framework for Real-Time Small Object Tracking via Slice-Assisted Training and Adaptive Association

2025-07-16

AI Summary

This paper addresses the challenging multi-object tracking (MOT) problem for small, agile targets (e.g., birds) under drone-captured imagery—characterized by sparse visual features, severe motion entanglement, and dense occlusions leading to identity ambiguity. To tackle these issues, we propose an efficient, robust, and real-time tracking framework. Methodologically: (1) We introduce SliceTrain, a novel training augmentation strategy combining deterministic full-coverage slicing with stochastic augmentation to enhance small-object detection robustness; (2) We design an appearance-agnostic association mechanism incorporating motion direction persistence and adaptive similarity measurement to mitigate occlusion-induced identity switches and motion confusion; (3) We extend YOLOv8+OC-SORT with bounding-box expansion and distance-aware penalty in the association stage. Evaluated on the SMOT4SB benchmark, our method achieves a state-of-the-art SO-HOTA score of 55.205, demonstrating superior performance and practical viability in real-world complex aerial scenarios.

Technology Category

• Computer Vision: Motion & Tracking
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Security and Privacy: Tracking, profiling, and countermeasures against them

Abstract

Tracking small, agile multi-objects (SMOT), such as birds, from an Unmanned Aerial Vehicle (UAV) perspective is a highly challenging computer vision task. The difficulty stems from three main sources: the extreme scarcity of target appearance features, the complex motion entanglement caused by the combined dynamics of the camera and the targets themselves, and the frequent occlusions and identity ambiguity arising from dense flocking behavior. This paper details our championship-winning solution in the MVA 2025 "Finding Birds" Small Multi-Object Tracking Challenge (SMOT4SB), which adopts the tracking-by-detection paradigm with targeted innovations at both the detection and association levels. On the detection side, we propose a systematic training enhancement framework named extbf{SliceTrain}. This framework, through the synergy of 'deterministic full-coverage slicing' and 'slice-level stochastic augmentation, effectively addresses the problem of insufficient learning for small objects in high-resolution image training. On the tracking side, we designed a robust tracker that is completely independent of appearance information. By integrating a extbf{motion direction maintenance (EMA)} mechanism and an extbf{adaptive similarity metric} combining extbf{bounding box expansion and distance penalty} into the OC-SORT framework, our tracker can stably handle irregular motion and maintain target identities. Our method achieves state-of-the-art performance on the SMOT4SB public test set, reaching an SO-HOTA score of extbf{55.205}, which fully validates the effectiveness and advancement of our framework in solving complex real-world SMOT problems. The source code will be made available at https://github.com/Salvatore-Love/YOLOv8-SMOT.

Problem

• Tracking small agile objects from UAV views
• Addressing scarce features and motion complexity
• Solving occlusion and identity ambiguity in flocks

Innovation

• SliceTrain enhances small object detection training
• Motion direction maintenance ensures stable tracking
• Adaptive similarity metric handles irregular motion