A hybrid opto-atomic spatio-temporal holographic correlator performs 3D convolutions for video action recognition, achieving 59.72% accuracy on a four-class dataset with projected operation at 125,000 frames per second.
Quo Vadis, action recognition? A new model and the kinetics dataset,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AR 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Opto-Atomic Spatio-Temporal Holographic Correlators for High-Speed 3D CNNs
A hybrid opto-atomic spatio-temporal holographic correlator performs 3D convolutions for video action recognition, achieving 59.72% accuracy on a four-class dataset with projected operation at 125,000 frames per second.