TSN feature on ActivityNet v1.3. The frame interval is set to 16. Using linear interpolation, each video feature sequence is rescaled to L = 100 snippets.
Supplementary material of the paper "The power of deep without going deep? A study of HDPGMM music representation learning" ====== Authors: Jaehun Kim (jaehun.j.kim@gmail.com) Cynthia C.S. Liem # Gene