5 merge paired plane registration into this capsule #8
Conversation
code/paired_plane_registration.py
Outdated
| @@ -195,16 +202,18 @@ def paired_plane_cached_movie(h5_file: Path, | |||
| assert len(data_length) == len(y_shifts) == len(x_shifts) | |||
| if run_nonrigid: | |||
| assert len(data_length) == ymax1.shape[0] == xmax1.shape[0] | |||
There was a problem hiding this comment.
this line can go up after L192
| f.create_dataset('data', data=r_frames) | ||
| if i == 0: | ||
| with h5py.File(temp_path, 'w') as f: | ||
| f.create_dataset('data', (data_length, 512, 512), maxshape=(chunk_size, 512, 512), chunk=(1000, 512, 512)) |
There was a problem hiding this comment.
I think it is better this way to fix data shape before saving the data, so that we don't need to check later.
I think the first chunk is not saved. (maybe just need to remove else: in L228?
Hardcoding 512 x 512 is not safe- let's use data shape (from registered.h5 file).
Is chunks (not chunk i guess..?) (1000, 512, 512) arbitrary, or is it because of emf? for EMF we'll access evenly distributed 1000 epochs so having 1000 chunks.shape[0] may not be efficient.
| oeid_pj = list(input_dir.glob(f"{oeid}_processing.json"))[0] | ||
| with open(oeid_pj, "r") as f: | ||
| pj = json.load(f) | ||
| if "nonrigid" in pj["data_processes"][0]["parameters"]["nonrigid"]: |
There was a problem hiding this comment.
Can you check if 'nonrigid' is recorded as 'nonrigid'/'rigid' or 'True'/'False'?
There was a problem hiding this comment.
@jkim0731 I did check and this is directly from the output parameters from suite2p
|
****Closing to merge into development branch |
arielleleon
deleted the
5-merge-paired-plane-registration-into-this-capsule
branch
@jkim0731