zddhub/opensse

Runtime error

zddhub opened this issue · 14 comments

Reported by @warmhug

Follow the doc of https://github.com/zddhub/opensse/wiki/How-to-train-data , when I run extract command, it throws errors:

$ sse extract -d ~/Database/SHREC12/ -f ~/Database/SHREC12/2012_filelist -o ~/Database/SHREC12/2012_
dyld: Library not loaded: @rpath/libopencv_core.3.2.dylib
  Referenced from: /usr/local/bin/extract
  Reason: image not found

My installation directory are /usr/local/opt/opencv3/include(actually in/usr/local/Cellar/opencv3/HEAD-c1007c7/include) and /usr/local/Cellar/boost/1.63.0/include/boost, they are homebrew's directory.

Is this error because my installation directory is not appropriate?

Please add opencv library (like as libopencv_core.3.2.dylib) to your library path

Done. But may encounter some new mistakes, the content of 2012_keypoints and 2012_features file are all just 0 and no more. The contents of the 2012_filelist file are empty.

$ sse extract -d ~/Database/SHREC12/ -f ~/Database/SHREC12/2012_filelist -o ~/Database/SHREC12/2012_
Extract descriptors 0/0.

And

$ sse vocabulary -f ~/Database/SHREC12/2012_features -n 1000 -o ~/Database/SHREC12/vocabulary
Assertion failed: (collection.size() >= numclusters), function kmeans_init_random, file /Users/hua/inner/opensse/../opensse/sse/vocabulary/kmeans_init.h, line 30.
cluster ...

So sad! I will try that and make it work

Please give sse extract more times (~30+ minutes ) to run, when I run it, and use tail -f 2012_features as below:

screenshot

It means feature are extracting now.

I will print prompt info when exacting

sse extract end Immediately. Not as your gif picture shows.

Do you download the database and get 102 line drawing views for each model?

Sorry, I forgot, and I get it, but the document did not say how to use it?

I think i have written all things, please help update if it is not clear

image

The question is how to use "line drawing views" with these sse filelist / extract commands?

This is another project, you can go to its README file and use gen_view_image command. This command only projects one modal to 102 views images each time, perhaps you need a script to deal with more modals

I follow the README file, but they have nothing to do with sse .. command. So I can't understand also. like sse filelist -d ~/Database/SHREC12/ -p "*.jpg" -o ~/Database/SHREC12/2012_filelist Where the "*.jpg" file are?

I have another project to project 102 views images for each model, please link to https://github.com/zddhub/trianglemesh for detail.

gen_view_image is not a sub-command of sse, its usage:
gen_view_image modelfile xfdir viewnum imagedir

How's going on? Did you fix this bug?

Now it works fine, you can close this issue, thanks