Knowledge Distillation in Fourier Frequency Domain for Dense Prediction

Knowledge distillation has been widely used to enhance student network performance for dense prediction tasks. Most previous knowledge distillation methods focus on valuable regions of the feature map in the spatial domain, ignoring the semantic information in the frequency domain. This work explore...

Ausführliche Beschreibung

Bibliographische Detailangaben

Veröffentlicht in:	IEEE signal processing letters 2025, Vol.32, p.296-300
Hauptverfasser:	Shi, Min, Zheng, Chengkun, Yi, Qingming, Weng, Jian, Luo, Aiwen
Format:	Artikel
Sprache:	eng
Schlagworte:	Accuracy Dense prediction Detectors Discrete Fourier transforms Feature extraction Fourier transform Frequency-domain analysis Head knowledge distillation Knowledge engineering Object detection semantic segmentation Semantics Technological innovation
Online-Zugang:	Volltext bestellen
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!