Ego4D v2 test set
收藏arXiv2025-09-30 收录
下载链接:
https://github.com/CorrineQiu/Ego4D-LTA-Challenge-2025
下载链接
链接失效反馈官方服务:
资源简介:
该数据集旨在评估基于编辑距离度量的长期动作预期任务的性能,它衡量预测的动作序列与真实动作序列的准确性。此外,该数据集还用于长期动作预期任务的公共排行榜,其中性能评估采用编辑距离(ED)指标。
This dataset is designed to evaluate the performance of long-term action anticipation tasks based on edit distance metrics, as it measures the accuracy between predicted action sequences and ground-truth action sequences. Furthermore, this dataset is also utilized for the public leaderboard of long-term action anticipation tasks, where performance evaluation adopts the edit distance (ED) metric.
提供机构:
Ego4D



