HiCE with grasp-aware loss and soft labels outperforms baselines by 16.91% AP on the new TouchMoment dataset for frame-level hand contact detection in first-person video.
Grounded language-image pre-training
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
-
Detecting Precise Hand Touch Moments in Egocentric Video
HiCE with grasp-aware loss and soft labels outperforms baselines by 16.91% AP on the new TouchMoment dataset for frame-level hand contact detection in first-person video.