Number of output estimation images
holistudio opened this issue · comments
More of a question than an issue: I notice that for a new set of 78 action images, create_movie.py ends up doing pose estimation for only the first 75 images. What causes this difference?
The most I could deduce from the code so far is that tf.app.flags.DEFINE_integer("seqlen", 5, "Length of sequence")
in create_movie.py which truncates the input data at n_extra = n2d%FLAGS.seqlen
How is seqlen defined?
This is not urgent, but any guidance on this would be appreciated.