Closed fengyang0317 closed 8 years ago
Hi fengyang0317, are you still using this code? I'm having difficulty in using the heatmap-Caffe, and a lot of strange errors came up when I want to run the .sh file. Could you tell me the version of your CUDA? I want to know if that matters. Thx!
Hi Feng, The default value of multfact should be 1 (it is a multiplication factor for the regression coordinates in your label file). This has been fixed in the latest commit.
I'm having the same error please; and I couldn't find a solution. What I want to know is how to set the value of "multfact" please ?? because in the readme file, is said that "Note these files require multfact=282 in both training and testing data layers": mean that the images have to have mulfact=282 and @tpfister you said that the dafault value is 1? Can you explain me that please ;
Hi, If you are using Pfister's data, you should set multfact to 282. If you prepare data yourself, it should be 1.
Pfister's data is transformed using the value 282.
I prefer to use fengyang0317@gmail.com.
Yang
At 2016-02-19 21:59:50, "Sameh Neili" notifications@github.com wrote:
I'm having the same error please; and I couldn't find a solution. What I want to know is how to set the value of "multfact" please ?? because in the readme file, is said that "Note these files require multfact=282 in both training and testing data layers": mean that the images have to have mulfact=282 and @tpfister you said that the dafault value is 1? Can you explain me that please ;
— Reply to this email directly or view it on GitHub.
Thanks @fengyang0317. So I have to add this variable "multfact" in the "train_val.prototxt" file , and set it to 282, isn't ?
yes,if you want to use his data.
在 2016-02-23 17:40:47,"Sameh Neili" notifications@github.com 写道:
Thanks @fengyang0317. So I have to add this variable "multfact" in the "train_val.prototxt" file , and set it to 282, isn't ?
— Reply to this email directly or view it on GitHub.
@xuxy09 did you finally succed to compile and run this code of caffeHeatmap? I still have many bugs and don"t know how to solve that , if you could post your errors and how did you solve them ? pearhaps that could help me and some others freinds as well :)
Multfact has been addressed in the latest commit
The value is 0 by default. However, it is multiplied and divided by some some other value in data_heatmap.cpp, leading to nan error.