DyFrDet: Towards Accurate Small Object Detection via Dynamic Frequency Suppression with Label Disambiguation

Researchers unveiled DyFrDet, a small-object detector designed to boost accuracy with dynamic frequency suppression and label disambiguation.

Its dynamic frequency-aware feature pyramid trims redundant low-frequency content and high-frequency noise, improving localization precision.

A label disambiguation module helps the model handle ambiguous targets, a common challenge in small-object detection.

The method reports state-of-the-art results across multiple benchmarks, suggesting strong potential for real-world vision tasks.