search

qianyuzqy / TransVOD_Lite

(TPAMI 2023) TransVOD:End-to-End Video Object Detection with Spatial-Temporal Transformers (implementations of TransVOD Lite).
Apache License 2.0
37 stars 6 forks source link

数据集请教 #21

Open zzhangfan opened 7 months ago

zzhangfan commented 7 months ago

嗨！非常感谢您将如此优秀的工作进行了开源。我有一个训练集构造的疑惑点，向您请教。请问clip是基于每个视频段先进行15帧挑选，然后在15帧中进行构造？还是基于原始的视频段？