Yan Ling

Results 57 comments of Yan Ling

您好,Faster RCNN抽取的各个目标区域框,不是切分的patch,模型本身会输出抽取的区域狂的类别概率分步。

参考代码出处https://github.com/jiasenlu/bottom-up-attention/blob/master/tools/generate_tsv.py

是的,但需要在运行faster rcnn的时候添加进类别概率这一项

> 那个区域框 和 语义类别的概率分布 这些数据在哪个文件夹里,叫啥名 ![image](https://user-images.githubusercontent.com/63391363/228861107-a456db78-0cb9-45a1-ae22-5264b39f458d.png) 在region box里

直接使用pip install fastnlp安装即可,我们当时的版本是fastnlp==0.7.0

This error seems the model can't find the initial bart model path, you can download the bart model files from hugging face and modify the path.

![image](https://user-images.githubusercontent.com/32841247/211474768-e0f46cfa-69e8-4cab-96d0-21c22dcab7b9.png) you need to change the parameter "pretrained_model_name" as the path of your download bart model.

在google云盘的压缩包里有提供