CI-ICM optimizes learned image coding for machines via channel importance modules, delivering 16.25% and 13.72% BD-mAP gains on detection and segmentation over baseline.
SC2 benchmark: Supervised compression for split computing,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.IV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
CI-ICM: Channel Importance-driven Learned Image Coding for Machines
CI-ICM optimizes learned image coding for machines via channel importance modules, delivering 16.25% and 13.72% BD-mAP gains on detection and segmentation over baseline.