1 paper · 1 filter
Snehal Singh Tomar, Alexandros Graikos, Arjun Krishna +2
Modern deep learning methods typically treat image sequences as large tensors of sequentially stacked frames. However, is this straightforward representation ideal given the curren…