SyncVSR (Word Boundary)

Reported on 6 benchmarks across 2 tasks · 1 paper · 6 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Computer Vision3 results

LipreadingonCAS-VSR-W1k (LRW-1000)
Top-1 Accuracy· 2024-06-18
58.2
SOTA
SyncVSR: Data-Efficient Visual Speech Recognition with End-to-End Crossmodal Audio Token Synchronization arXiv:2406.12233
LipreadingonLip Reading in the Wild
Top-1 Accuracy· 2024-06-18
95
SOTA
SyncVSR: Data-Efficient Visual Speech Recognition with End-to-End Crossmodal Audio Token Synchronization arXiv:2406.12233
LipreadingonLRW
Top 1 Accuracy· 2024-06-18
80.3
SOTA
SyncVSR: Data-Efficient Visual Speech Recognition with End-to-End Crossmodal Audio Token Synchronization arXiv:2406.12233

Natural Language Processing3 results

Natural Language TransductiononCAS-VSR-W1k (LRW-1000)
Top-1 Accuracy· 2024-06-18
58.2
SOTA
SyncVSR: Data-Efficient Visual Speech Recognition with End-to-End Crossmodal Audio Token Synchronization arXiv:2406.12233
Natural Language TransductiononLip Reading in the Wild
Top-1 Accuracy· 2024-06-18
95
SOTA
SyncVSR: Data-Efficient Visual Speech Recognition with End-to-End Crossmodal Audio Token Synchronization arXiv:2406.12233
Natural Language TransductiononLRW
Top 1 Accuracy· 2024-06-18
80.3
SOTA
SyncVSR: Data-Efficient Visual Speech Recognition with End-to-End Crossmodal Audio Token Synchronization arXiv:2406.12233