M6Doc
M6Doc copied to clipboard
Thank you for your excellent work. I have a question here. Is the mAP indicator in the paper tested on the evaluation set or the test set?
Thanks a lot for this contribution. It would be appreciated to also share the trained models from the paper or at least the high-performance ones such as ISTR. Best, Omar
数据集链接不可用
The two download links are invalid.
您好!请问是否有一套可供使用的对于该数据集的评估代码,以实现得出其他的模型在此数据集上的表现,因为我自己在使用该数据集进行评估时,遇到了很多问题,比如是否应该需要使用预测框和标注框首先根据某种规则进行匹配,然后再计算相关指标?如果是预测框是多个小框(比如一个段落中的多句话)而标注框是大框(比如一个大段落)时应该怎么计算指标,反之如果是一对多呢?
I requested non-commercial access to the dataset two months ago, but have not received a response. I have sent the request again, please help to confirm the request.
Your team's work is very good. Can you exchange links in realme? I have developed a UniOne dataset aimed at solving the problem of document parsing upstream and downstream task...