LVTP prunes transformer tokens during semantic segmentation using multi-scale Tsallis entropy with Sobel edge guidance, reporting 20-46% FLOP reduction with a few mIoU points lost and no retraining.
Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Back to Fundamentals: Low-Level Visual Features Guided Progressive Token Pruning
LVTP prunes transformer tokens during semantic segmentation using multi-scale Tsallis entropy with Sobel edge guidance, reporting 20-46% FLOP reduction with a few mIoU points lost and no retraining.