error in cnn_ucf101_get_im_flow_batch.m

[ixhorse]

When I run cnn_ucf101_fusion.m, I get the error about
"Subscript indices must either be real positive integers or logicals"
at the code
"last_frame = min(frameSamples(end), max(nFrames - nStack/4 - opts.nFrameStack,nStack/4 ));"
in file "cnn_ucf101_get_im_flow_batch.m".
It seems "framSamples" is empty. Does it mean that I didn't load the data successfully?

[onegoldensun]

have you solved it? i have the same problem

[ixhorse]

It is the path error. "data" is your data path, RGB and flow datasets should in the path 'data/data/ucf101/tvl1_flow' and 'data/ucf101/jpegs_256'.
The codes in cnn_setup_environment.m is here.
opts.flowDir = 'data/ucf101/tvl1_flow';
opts.imageDir = 'data/ucf101/jpegs_256';
Replace the 'data' with your data path and do not change other path in these two lines.
@onegoldensun

[Victoria2333]

@ixhorse do you meet errors running cnn_ucf101_spatial.m,I have already changed my datapath,and there is some other errors:
No public property dilate exists for class dagnn.Conv
Error in dagnn.Layer/load(line 191)
obj.(f)=s.(f)
error in dagnn.dagnn.loadobj(lin27)
block.load(struct(s.layers(l).block))
error in cnn_ucf101_spatial(line112)
net=dagnn.DagNN.loadobj(net)
Another question:And if I train on my own dataset,do I only need to use cnn_ucf101_spatial.m,cnn_ucf101_temporal.m and cnn_ucf101_fusion.m??Do i need to change some configuration in other .m files?
Thanks for you attention

[lvyali]

It is the path error. "data" is your data path, RGB and flow datasets should in the path 'data/data/ucf101/tvl1_flow' and 'data/ucf101/jpegs_256'.
The codes in cnn_setup_environment.m is here.
opts.flowDir = 'data/ucf101/tvl1_flow';
opts.imageDir = 'data/ucf101/jpegs_256';
Replace the 'data' with your data path and do not change other path in these two lines.
@onegoldensun

what about others codes in cnn-setup_environment.m?Don't change anything except these two lines?

[ZYF131]

Hi, have you successfully run this code?