BEM is a training-free inference module that suppresses false positives in fixed-background object detection by maintaining background embedding prototypes and applying an inverse-similarity rank-weighted penalty to detection logits.
In: International Conference on Neural Information Processing
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
BEM: Training-Free Background Embedding Memory for False-Positive Suppression in Real-Time Fixed-Background Camera
BEM is a training-free inference module that suppresses false positives in fixed-background object detection by maintaining background embedding prototypes and applying an inverse-similarity rank-weighted penalty to detection logits.