RESEARCH RECORD / AnyTemplate
METHOD / FINDINGS
A zero-shot template-matching framework that combines visual foundation models, aspect-ratio-preserving normalization, cross-attention, and geometric refinement across image scales.
The geometry-aware normalization preserves aspect ratio and pads to the encoder canvas. A trainable cross-attention interaction module exchanges context between template and proposal features.
94.1% accuracy on DEAMPB. Zero-shot template matching evaluation; use accuracy, not mAP, when citing Table II. (Table II, p. 5.)
A lightweight PCB defect detector that preserves shallow high-resolution features through multi-scale adaptation, semantic alignment, and wavelet enhancement.
Specialized edge, wavelet, texture, and directional filters form an expert ensemble that detects subtle forgery traces in remote-sensing imagery.
A zero-shot template-matching framework that combines visual foundation models, aspect-ratio-preserving normalization, cross-attention, and geometric refinement across image scales.
