Conversation
|
Verified that XML output of split_dataset matches xml from bigstitcher's Virtual Split command. |
| timepoint = il.get("timepoint") | ||
| file_path = il.find("path").text if il.find("path") is not None else None | ||
| channel = file_path.split("_ch_", 1)[1].split(".ome.zarr", 1)[0] | ||
| file_path = il.get("path") |
There was a problem hiding this comment.
Update parse_image_loader_zarr() to properly handle bigstitcher xml conventions, and channel default when not in filename
| if vid == view_id: | ||
| to_process_interval = (lb, ub) | ||
|
|
||
| ub_inclusive = (ub[0]+1, ub[1]+1, ub[2]+1) |
There was a problem hiding this comment.
interval needs to use inclusive coordinates for upper bound, so that it is compatible with block_interval coordinates when calling self.contain()
| return int(a + b - (a % b)) | ||
|
|
||
| def find_min_step_size(self): | ||
| def find_min_step_size(self, lowest_resolution=(1.0, 1.0, 1.0)): |
There was a problem hiding this comment.
Update this to allow for blocks to be created at 1.0 pixel resolution (previously was limited to blocksizes of powers of 64)
| outer_timepoints = ET.Element('Timepoints', {'type': 'pattern'}) | ||
| ip = ET.SubElement(outer_timepoints, 'integerpattern') | ||
| ip.text = "0" | ||
| tps = sorted({int(v['old_view'][0]) for v in self.self_definition if v['old_view'][0] is not None}) |
There was a problem hiding this comment.
Determine timepoints integer range from pre-split views
|
|
||
| class SplitImages: | ||
| def __init__(self, target_image_size, target_overlap, min_step_size, data_gloabl, n5_path, point_density, min_points, max_points, | ||
| def __init__(self, target_image_size, target_overlap, min_step_size, data_global, n5_path, point_density, min_points, max_points, |
There was a problem hiding this comment.
fix typo in data_global
| if size < 0: | ||
| size = l + size | ||
| return size | ||
| def last_image_size(self, L, S, O): |
There was a problem hiding this comment.
Rewrite of this function to properly determine the size of the last tile. Logic was incorrect previously. perfect uniform tiling has been verified
|
|
||
| if length <= self.target_image_size[i]: | ||
| pass | ||
| dim_intervals.append((0, length - 1)) |
There was a problem hiding this comment.
if the entire dataset is smaller than blocksize, still make one block - use the size of the dataset in this dimension
| for j in range(i): | ||
| other_interval = intervals[j] | ||
| intersection = self.intersect(interval, other_interval) | ||
| new_v_ip_l = [] |
There was a problem hiding this comment.
Large diff starting here changes handling of IP detections when splitting:
- adds support for not creating any fake interest points when splitting.
- adds support for virtual splitting of tiles when no IPs have been detected.
| file_path = il.find("path").text if il.find("path") is not None else None | ||
|
|
||
| channel = file_path.split("_ch_", 1)[1].split(".ome.zarr", 1)[0] | ||
| timepoint = il.get("tp") or il.get("timepoint") |
There was a problem hiding this comment.
Update xml_to_dataframe_split.parse_image_loader_zarr() to match changes to data_prep.xml_to_dataframe.parse_image_loader_zarr()
|
test fusion by running split, then calling fusion to test that it fusion works. run rhapso pipeline all the way through
|
1 participant
Description of the Changes You Made
Type of Change
Instructions for Testing
Additional Info