happynear / PainRegression

Transferring Face Verification Nets To Pain and Expression Regression
MIT License
39 stars 18 forks source link

matlab crashes when running ./scripts/extract_feature.m #1

Open tottitt opened 6 years ago

tottitt commented 6 years ago

when I run ./scripts/extract_feature.m, I got the following error messages:

I1119 12:20:25.223507 9956 layer_factory.hpp:77] Creating layer data I1119 12:20:25.223526 9956 net.cpp:84] Creating Layer data I1119 12:20:25.223532 9956 net.cpp:380] data -> data I1119 12:20:25.223543 9956 net.cpp:380] data -> label I1119 12:20:25.223556 9956 image_data_layer.cpp:39] Opening file /home/tt/TOSHIBA/deepFaceStuff/PainRegression/prototxt/sublist/1/list_val.txt I1119 12:20:25.562822 9956 image_data_layer.cpp:107] A total of 2134 images. E1119 12:20:25.563374 9956 io.cpp:91] Could not open or find file I1119 12:20:25.223507 9956 layer_factory.hpp:77] Creating layer data I1119 12:20:25.223526 9956 net.cpp:84] Creating Layer data I1119 12:20:25.223532 9956 net.cpp:380] data -> data I1119 12:20:25.223543 9956 net.cpp:380] data -> label I1119 12:20:25.223556 9956 image_data_layer.cpp:39] Opening file /home/tt/TOSHIBA/deepFaceStuff/PainRegression/prototxt/sublist/1/list_val.txt I1119 12:20:25.562822 9956 image_data_layer.cpp:107] A total of 2134 images. E1119 12:20:25.563374 9956 io.cpp:91] Could not open or find file /home/tt/TOSHIBA/deepFaceStuff/Data/aligned-Images042-ll042/ll042t1aaaff/ll042t1aaaff001.png F1119 12:20:25.563396 9956 image_data_layer.cpp:127] Check failed: cv_img.data Could not load /home/tt/TOSHIBA/deepFaceStuff/Data/aligned-Images042-ll042/ll042t1aaaff/ll042t1aaaff001.png Check failure stack trace:

F1119 12:20:25.563396 9956 image_data_layer.cpp:127] Check failed: cv_img.data Could not load /home/tt/TOSHIBA/deepFaceStuff/Data/aligned-Images042-ll042/ll042t1aaaff/ll042t1aaaff001.png Check failure stack trace:

how ever /TOSHIBA/deepFaceStuff/Data/aligned-Images042-ll042/ll042t1aaaff/ll042t1aaaff001.png is just there

new_height and new_width are both zero before calling

cv::Mat cv_img = ReadImageToCVMat(rootfolder + lines[linesid].first, new_height, new_width, is_color);

in image_data_layer.cpp

eglxiang commented 6 years ago

The images are successfully read but the new size are not right.

happynear commented 6 years ago

If you don't want to change the resolution of the image, then new_height and new_width is ok to be zero.

applefishsky009 commented 6 years ago

I want to have a try on EmotioNet and confused about the input, but EmotionCatFile-train.txt & EmotionCatFile-test.txt & FACSFile-weight-train.txt & FACSFile-weight-test.txt is missed in this repo. Could you please help me figure it out?