One-sentence Summary: First single-model anytime approach for pixel-level visual recognition; our model with redesigned exits and confidence adaptivity enables ...
確定! 回上一頁