vame.csv_to_npy pose confidence is hard coded as 0.99
stowerslab opened this issue · 1 comments
stowerslab commented
i was wondering why vame.create_trainset() was throwing empty array errors and found out that one of my columns in the NPY file was still empty even when i lowered the pose confidence threshold
i think line 5 in the code block below is the issue
for i in range(bodyparts*2):
if i % 2 == 0:
idx +=1
seq = body_position[:,i]
seq[con_arr[idx,:]<.99] = np.NaN
body_position_nan.append(seq)
kvnlxm commented
Thank you for bringing this up. The csv_to_numpy.py
function is now updated and creates now the correct representation to create the training set and train VAME.
Cheers,
Kevin