Combined techniques reduce drone small object detection accuracy
Small Object Detection in Drone Aerial Imagery with LAF-YOLOv10
Computer Vision and Pattern Recognition
Summary
Detecting tiny objects in images taken by drones is tricky because the objects look very small and drones have limited computer power. The authors tried combining four methods that each seemed to help on their own, but when put together, the results were worse instead of better. They found that one specific change, swapping detection parts and how they interacted, caused most of the accuracy drop. This means we cannot assume improvements will always add up nicely when combined, and need to test combinations carefully.
What this means in practice
- •For drone developers: Identify the impact of combining small object detection methods on drone hardware to better design efficient models for aerial surveillance.
- •For autonomous vehicle engineers: Understand limitations of small object detection methods when transferring UAV-trained models to similar tasks, such as ground vehicle object spotting.
Authors
Quratulain Nayeem, Fahmina Taranum, Mohammed Mudassir Uddin
Abstract
General-purpose object detectors lose accuracy on UAV footage, where targets span only a handful of pixels and onboard compute is limited. Prior work composes independently-validated architectural techniques into one detector, assuming gains reported in isolation transfer once combined. We stress-test that assumption directly. LAF-YOLOv10 integrates four techniques into YOLOv10n: a Partial Convolution C2f (PC-C2f) backbone block, an Attention-Guided Feature Pyramid Network (AG-FPN), a P2 detection head replacing the large-object P5 head, and Wise-IoU v3 regression, asking whether their combined effect matches what each contributes alone. We train LAF-YOLOv10 three times (seeds 42, 123, 256) on VisDrone-DET2019, benchmark against unmodified YOLOv10n, and use TIDE error decomposition, per-category breakdown, per-component ablation, attention/loss comparisons, zero-shot transfer to UAVDT, and held-out/test-dev evaluation to localize where the combination succeeds or fails. Composability does not hold here. LAF-YOLOv10 reaches 24.0+/-0.4% mAP@0.5 at 2.14M parameters, 7.8 points below YOLOv10n (31.8%), a deficit that transfers to UAVDT (-10.0 points) and is confirmed by held-out and test-dev evaluation (23.5%, 22.5%). Background false positives, localization error, and duplicate detections move in the direction AG-FPN and Wise-IoU were designed to push. Ablation traces the deficit to a specific source: the P2/-P5 head swap costs 2.5 points independently plus a 2.5-point interaction penalty when layered onto a backbone weakened by PC-C2f, whose own 2.0-point loss is consistent with a partial pretrained-weight transplant (73/150 backbone tensors transfer). The failure is attributable to a specific interaction, not the four components individually. Composability must be verified directly, not assumed. Code/checkpoints: https://github.com/Mudassiruddin7/Small-Object-Detection-in-UAV-Imagery.