poddo

Results 45 comments of poddo

> 你好,我想问下您可不可以给我一份如何实现计算召回率的代码实现,小白入门时间不长,希望得到您的指导,谢谢啦! 你好,请问下可以看下怎么求精确度和召回率的代码呀,谢谢。

请问下你是怎么测试每个网络的precision和recall值的啊,感激不尽。

![image](https://user-images.githubusercontent.com/12837585/62689062-a608c300-b9fc-11e9-814e-85e909385377.png) 在MTCNN_model下新建RNet_No_Landmark和ONet_No_Landmark,并把RNet_landmark和ONet_landmark下的模型分别拷贝到上面新建的文件夹中,即可运行。

在gen_landmark_aug_24.py的62行的data = getDataFromTxt(ftxt)修改为: data = getDataFromTxt(ftxt,data_path=data_path) 并在前面添加data_path = '../../DATA/train'

wider_face_train_bbx_gt.txt格式具体内容参考如下: File name Number of bounding box x1, y1, w, h, blur, expression, illumination, invalid, occlusion, pose

![image](https://user-images.githubusercontent.com/12837585/62852231-6a7f3900-bd1b-11e9-9b08-ebe67f15fe6f.png) 首先贴一张我修改代码后的截图,注意这的name有四个值,每一个值运行一次,共四次。对照这gen_RNet_tfrecords.py文件来看,可以看出gen_ONet_tfrecords.py文件是少了些东西的,我是一步一步对照着gen_RNet_tfrecords.py文件来改的,基本上流程和代码是一样的。

my environment is (tensorflow 1.14.0, Cuda 9.0), and it's training well.

![image](https://user-images.githubusercontent.com/12837585/62613295-c6247d80-b93b-11e9-90e9-6ab44351afc6.png) I get this error when I run this command:python script.py –g=gt.zip –s=submit_nvlink_0.zip. I check the img_100.txt and it's ok, have someone met this problem or known what is the...

> > > > ![image](https://user-images.githubusercontent.com/12837585/62613295-c6247d80-b93b-11e9-90e9-6ab44351afc6.png) > > I get this error when I run this command:python script.py –g=gt.zip –s=submit_nvlink_0.zip. > > I check the img_100.txt and it's ok, have someone...