Closed Iagea closed 1 year ago
Hi @Iagea ,
Why did the error occur in the first place and which document caused the error?
Hi @Kohulan,
any PDF document I've tried (4) when running directly the script segment_structures_in_document.py
as well as when running directly the example Notebook with the example test_page.pdf when expand = True
for the segment_chemical_structures
function:
segments = segment_chemical_structures(np.array(pages[0]), expand=True, visualization=True)
PS: Tensorflow version 2.11.0.
Could you reproduce the same error using Tensorflow 2.10.0 or 2.12.0?
Ok, I just tried with two different environments and no, Tensorflow 2.10.0 and 2.12.0 also works fine. Must be a Tensorflow 2.11.1 problem.
Forget about it then
Thanks for confirming this issue. I would for now stick with TF 2.10.
Hi,
I was getting an error on line 451 from
decimer_segmentation/complete_structure.py
:arrays to stack must be passed as a "sequence" type such as list or tuple.
It's fixed changing the line 451 to
mask_array = np.stack(list(expanded_split_mask_arrays), -1)
Best, Isa.