3 papers
cs.CV2025
Progressive Class-level Distillation
Jiayan Li, Jun Li, Zhourui Zhang +1
In knowledge distillation (KD), logit distillation (LD) aims to transfer class-level knowledge from a more powerful teacher network to a small student model via accurate teacher-st…
cs.CV2025
SAMKD: Spatial-aware Adaptive Masking Knowledge Distillation for Object Detection
Zhourui Zhang, Jun Li, Jiayan Li +1
Most of recent attention-guided feature masking distillation methods perform knowledge transfer via global teacher attention maps without delving into fine-grained clues. Instead,…
cs.CV2024
DFMSD: Dual Feature Masking Stage-wise Knowledge Distillation for Object Detection
Zhourui Zhang, Jun Li, Zhijian Wu +2
In recent years, current mainstream feature masking distillation methods mainly function by reconstructing selectively masked regions of a student network from the feature maps of…