A landmark-based sign language recognizer with 3D-to-1D knowledge distillation and a text correction network achieves state-of-the-art WER on PHOENIX14/14T with a 12.93 MB quantized model.
C.; Koller, O.; Hadfield, S.; and Bowden, R
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CY 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
KD-MSLRT: Lightweight Sign Language Recognition Model Based on Mediapipe and 3D to 1D Knowledge Distillation
A landmark-based sign language recognizer with 3D-to-1D knowledge distillation and a text correction network achieves state-of-the-art WER on PHOENIX14/14T with a 12.93 MB quantized model.