Research
DERA: Detached Edge-Residual Adaptation for Prohibited item Detection
arXiv:2609.12411v1 Announce Type: new Abstract: Prohibited-item detection in X-ray imagery remains challenging due to object superposition, weak texture, and material clutter which obscure both semant
arXiv:2609.12411v1 Announce Type: new Abstract: Prohibited-item detection in X-ray imagery remains challenging due to object superposition, weak texture, and material clutter which obscure both semantic appearance and object boundaries. We propose extbf{DERA}, a extbf{D}etached extbf{E}dge-extbf{R}esidual extbf{A}daptation framework for prohibited item detection under X-ray imagery. DERA combines hierarchical visual features with a parallel pixel-difference edge pyramid and learns an object-specific boundary prior from training-time contours of the instance masks. The detached prior gates edge-sensitive features, which are injected into the early visual stages through residual heads. This staged design preserves the foundation detector at the start of adaptation, isolates boundary supervision from semantic feature learning, and restricts the final adaptation stage to only (14.7)K trainable parameters. Evaluated on PIDray, CLCXray, and STCray, DERA improves the baseline by extbf{3.1}, extbf{1.6}, and extbf{2.4} AP points, respectively.
Related
- Adaptive Spectrum-Aware Feature Disentangled Network for Small Object Detection
- DenseScout: Algorithm-System Co-design for Budgeted Tiny Object Selection on Edge Platforms
- RDANet: Relative Degradation Aware Network for Infrared Small Target Detection
Source: arXiv cs.CV | 2026-09-14