Peng.Zhou
Peng.Zhou
> Hi, > > Are you facing this issue with just fcn or all the models? Maybe try vgg-unet The network I used is fcn8 in`image-segmentation-keras/keras_segmentation/models/fcn.py` . Today,I will read...
> Hi, > > Are you facing this issue with just fcn or all the models? Maybe try vgg-unet Hi,bro, why your segnet use the Upsamping to decode? Tensorflow records...
I have the same problem with you , because I want to extract fc features 1024 like inceptionV3, so I change the efficientnet-b4' fc 1792 to 1024, It will be...
hello,everybody, I have solved this problem. It just caused by the version of pytorch. most of all use the pytorch>=0.4,but this version of torch is 0.3 or lower. here is...
FileNotFoundError: [Errno 2] No such file or directory: '/data/wider_face_split\\wider_face_val.mat'
腾讯的这个repo代码写的如此的垃圾，真给腾讯丢脸。。。
这些问题问的很好，因为现在很多边缘计算的设备只支持caffe，原因是因为caffe的inference是C++写的，虽然提供的python的训练接口，但是提供的接口最终还是对应着C++的代码。 以上已经解决了你的2,3,4的问题了，至于1的问题，由于不同框架用的一些基本操作略有不同，比如tf的pading与caffe的pading不一样，所以在写padding的时候一般都是写pading的底层代码，而不是直接引用对应框架的padding接口。不一定是在原框架下才能调最好，只要你能弄懂哪里的操作两个框架不一样，你就知道为什么某一框架比另一个框架训练出来的模型差的原因了。
I meet the same trouble with you, Wishing you give me a solution, Hoping you have nice day.
> @funquiz Hi, how do you get mAP 75.7? I just get 22mAP on VOC dataset Hi,man,I am same with you .