search

zhang-can / ECO-pytorch

PyTorch implementation for "ECO: Efficient Convolutional Network for Online Video Understanding", ECCV 2018
BSD 2-Clause "Simplified" License
188 stars 93 forks source link

Train和Test分成123部分有何意义呢 #59

Closed xjock closed 4 years ago

xjock commented 4 years ago

互相有重复，为什么不合并在一起？

stomachacheGE commented 4 years ago

因为数据集不是很大，分成三个split分别训练三个模型，在三个split分别评测，更具有统计意义。