-
Notifications
You must be signed in to change notification settings - Fork 102
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
question about run "run_demo.m" #5
Comments
seems not results returned from C3D. could you also post your log files from C3D/Caffe? |
The log file content is: But the file "extract_image_features.bin" is exist,and I check the the file path that should be no problem ,I don't know how to solve the problem, |
you should check your directory during execution. did you change to demo directory: cd ./demo/. before running the demo? could you print out your execution path in the mat file right before exec ./pred/loc/feature_extract.sh ? |
I find a strange problem,as long as I run the "run_demo.m",th file content in the "/demo/pre/loc/demo_list_test_prefix_localization.lst " and "/demo/pre/loc/demo_list_test_uniform16_localization.lst" will be empty,but the file exist.so I think the above problem is relevant this problem |
are demo_list_test_prefix_proposal.lst and demo_list_test_uniform16_proposal.lst also empty? otherwise localization list is empty is due to your proposal network returns nothing. |
I may not be state the question clear. |
as I mentioned in the above, the localization list is empty is exactly due to the proposal network returns nothing - basically no segment is predicted to likely contain action. So I was wondering whether demo_list_test_prefix_proposal.lst and demo_list_test_uniform16_proposal.lst are also empty and whether you can run the proposal network successfully. |
demo_list_test_prefix_proposal.lst and demo_list_test_uniform16_proposal.lst are not empty and generate "/pred/pro/seg_swin.mat" file. |
are you testing with the demo video or your own video - probably the model cannot detect any actions in your provided video... but in that case usually are very small confidence score... all zeros are really strange to me... |
sorrt to be a bothered ,I have another question about the run_demo.m, I don't understand the code in line(122-126): |
so prob is also empty or all zero? prob is the confidence score of being action for each segment. |
prob are all zeros |
I read the run_demo.m carefully ,there is no file generate in the /pred/pro/output,so the prob is empty,I don't understand the role of the "/pred/pro/output",I would appreciate for your answer. |
"/pred/pro/output" is used to store all features extracted by C3D (here confidence score fc8 values). so this indicates your proposal network (at least the feature extraction part) is not running correctly. I would suggest you, just in case that you are not, to get familiar with feature extraction using original C3D. |
@zhengshou ......................................................................................................................................... [----------] 10 tests from BenchmarkTest |
as I mentioned in previous emails with you, we did not modify test units and thus running test units will not pass for sure. as for your question about how to generate list files, please refer to samples in scnn/experiments/THUMOS14/ |
error occured when example is running v_BaseballPitch_g01_c01/000103.jpg |
@Priyanka154 did you found the solution ? |
@junaidwahid NOT YET, |
when I run the "run_demo.m" ,the following question is:
init sliding window startsinit sliding window done in 2.4115 s
generate proposal startsgenerate proposal list done in 1.06 s
run proposal network starts
run proposal network done in 0.11494 s
read proposal results startsread proposal results done in 0.21835 s
generate localization list starts
generate localization list done in 0.000827 s
run localization results starts
run localization results done in 0.011421 s
read localization results starts
Improper assignment with rectangular empty matrix.
Error in run_demo (line 165)
seg_swin(:,9) = a;
Tried many ways, but still no solution.I would appreciated for some advice,very thanks!
The text was updated successfully, but these errors were encountered: