AI RESEARCH

DFIR-DETR: Frequency-Domain Iterative Refinement and Dynamic Feature Aggregation for Small Object Detection

arXiv CS.LG

ArXi:2512.07078v3 Announce Type: replace-cross Small object detection in complex scenes exposes a fundamental tension in neural network design: backbone attention distributes computation uniformly regardless of content, pyramid necks inflate activation magnitudes during upsampling without norm compensation, and bottleneck convolutions progressively smooth high-frequency edge components through accumulated spatial filtering.