Ground Truth Protocol Guidelines

Tommy Macrina's Directives (August - September 2025)


Executive Summary

This document compiles protocol clarifications and best practices for ground truth segmentation tasks as directed by Tommy Macrina in the #putracers Slack channel between August 1 and September 3, 2025. All guidelines are referenced to specific Slack conversations for verification.


I. Mitochondria Segmentation Protocol

1.1 Individual Segmentation Required

Each mitochondrion must be assigned a separate segment ID and color. Do not paint all mitochondria with the same segment.

Reference: Message m10

1.2 Oversegmentation Directive

The original task description stating "do NOT oversegment mitochondria" was incorrect and should be disregarded. Mitochondria must be oversegmented individually.

Reference: Message m8

1.3 Membrane Handling

Do not paint mitochondrial membranes. Leave a visible border/gap around each mitochondrion to represent the membrane boundary.

References: Message m8, Message m15

II. Myelin Segmentation Protocol

2.1 Inner Boundary (Cell Interface)

No padding between myelin and the cell it ensheathes. The myelin segment should directly contact the inner cell.

Reference: Message m15

2.2 Myelin-to-Myelin Contact

No padding between adjacent myelin segments. Myelin segments may touch each other.

Reference: Message m15

2.3 Outer Boundary (Adjacent Cells)

Myelin should approach but not touch the membranes of neighboring cells. Maintain a small gap between myelin exterior and other cell membranes.

Reference: Message m15

III. Upsampled Image Best Practices

3.1 Two-Pass Labeling Approach

Recommended workflow:

  1. First pass: Paint annotations on the original image (thicker sections)
  2. Second pass: Copy labels onto the upsampled image and connect objects naturally
Reference: Message m2

3.2 Multi-Resolution Viewing

Working exclusively from upsampled images can be misleading. Always cross-reference with original image data.

References: Message m3, Message m4

3.3 Handling Artifacts

"Hallucinations" (artificial structures in upsampled data) should be ignored. Focus on anatomically sensible structures.

References: Message m1, Message m11

IV. Ambiguous Cases & Edge Cases

4.1 Uncertain Organelles

When uncertain whether an object is a mitochondrion (vs. peroxisome, lysosome, etc.), segment it as a mitochondrion. Objects can be removed during review if needed.

Reference: Message m9

4.2 EM Damage Regions

Follow team consensus for handling EM damage. Specific protocols to be determined based on damage type.

Reference: Message m17

V. Workflow & Quality Assurance

5.1 Mandatory Peer Review

Before marking tasks as finished, at least one other team member must inspect your work. Prioritize sharing areas where you struggled.

Reference: Message m13

5.2 Sharing Annotations for Review

Correct procedure in WebKnossos:

  1. Navigate to Tasks
  2. Search for tasks under the relevant project
  3. Click the plus symbol (+) for the task
  4. Select "Actions" button (right side)
  5. Choose "View"
References: Message m13, Message m14

VI. WebKnossos Tools & Techniques

6.1 Interpolation Tool

The volume interpolation feature is particularly useful for objects perpendicular to the sectioning plane in higher z-resolution data. May require manual cleanup.

Reference: Message m7

6.2 3D Mesh Visualization

Reference: Message m12

6.3 SAM2 AI Segmentation (Experimental)

New AI-assisted segmentation tools available on experimental server:

Reference: Message m18

Reference Messages

Below are the complete Slack message threads referenced throughout this document.


Standalone Messages by Tommy Macrina

m1 — Standalone Message

Channel: putracers
Date: 2025-08-05 19:38 UTC
Author: tmacrina
View in Slack ↑ Back to Summary
From my spot checks, you've all done well: just ignore them.

m2 — Standalone Message

Channel: putracers
Date: 2025-08-05 19:41 UTC
Author: tmacrina
View in Slack ↑ Back to Summary
I realize now that we should probably handle labeling this data in two passes: 1. paint the original image (thicker sections) 2. copy those labels onto the upsampled image (computationally thinner sections), and just connect the objects as "naturally" as possible

m3 — Standalone Message

Channel: putracers
Date: 2025-08-05 19:41 UTC
Author: tmacrina
View in Slack ↑ Back to Summary
I've found only working from the upsampled image to be misleading.

m4 — Standalone Message

Channel: putracers
Date: 2025-08-05 20:18 UTC
Author: tmacrina
View in Slack ↑ Back to Summary
Here's a <https://spelunker.cave-explorer.org/#!middleauth+https://global.daf-apis.com/nglstate/api/v1/5712377138905088|side-by-side of the original image and the upsampled image>.

m5 — Standalone Message

Channel: putracers
Date: 2025-08-05 20:41 UTC
Author: tmacrina
View in Slack ↑ Back to Summary
Open left sidebar and see if "volume" is flipped on?

m6 — Standalone Message

Channel: putracers
Date: 2025-08-05 20:41 UTC
Author: tmacrina
View in Slack ↑ Back to Summary
I see your segmentation still.

Threaded Discussions with Tommy Macrina

m7 — Thread in putracers

Date: 2025-08-01 13:34 UTC
View in Slack ↑ Back to Summary
Ryan (2025-08-01 13:34:20 UTC):
Oh that’s what that does. That would be useful with the larger, simpler segments we typically start off with, but the smaller neurites need more guidance. I agree with Jay that it might require clean up but I’ll give it a try
tmacrina(TOMMY) (2025-08-01 15:30:16 UTC):
Yeah, this has been our general experience, too.
tmacrina(TOMMY) (2025-08-01 15:31:38 UTC):
Because this image has a higher z-resolution, more objects that are pependicular to the sectioning plane don't change that much. So you might find more objects where interpolation is useful.
Ryan (2025-08-01 15:59:35 UTC):
Yea I tried it on a simple round segment with minimal movement between slices with a 2 slice gap and it was pretty perfect. But I tried 5-6 slices in between and it got a bit confused by the size change in the segment and under-colored. But I'm gonna keep playing with it and find the sweet spot

m8 — Thread in putracers

Date: 2025-08-01 15:25 UTC
View in Slack ↑ Back to Summary
Jay (2025-08-01 15:25:29 UTC):
@Tommy Macrina @Eric Are we definitely oversegmenting mitochondria? The task directions explicitly say not to.
tmacrina(TOMMY) (2025-08-01 15:26:57 UTC):
Yes, it was a mistake to assign that task description.
Jay (2025-08-01 15:27:16 UTC):
Ok, thanks for the confirmation, we'll oversegment!
Eric (2025-08-01 15:27:41 UTC):
Sorry for the confusion, just was getting to this
bun(TOMMY) (2025-08-01 15:27:46 UTC):
do we leave a boarder between mitos?
bun(TOMMY) (2025-08-01 15:27:57 UTC):
for membrane
tmacrina(TOMMY) (2025-08-01 15:28:20 UTC):
Yes, try not to paint the mitochondria's membrane.
bun (2025-08-01 15:30:37 UTC):
is there a way we can get Kisuk's tool that expands or reduces the segments so we can make the membrane unpainted while still using the fill tools?
tmacrina(TOMMY) (2025-08-05 19:43:23 UTC):
> Kisuk's tool that expands or reduces the segments This is some post-processing he does. Handling it manually for now is the right call, until we have more experience with this data.
tmacrina(TOMMY) (2025-08-05 19:43:50 UTC):
It'd be nice if webknossos made it easy to extend their toolkit...
bun (2025-08-15 17:52:50 UTC):
i made a little app with react which demos a rudimentary shape deforming tool resembling the desired extruding/interpolating function in webknossos. i think if we could implement something like this demo in webknossos we could increase ground truth tracing speed and reduce painting fatigue
bun(TOMMY) (2025-08-15 18:08:27 UTC):
this way if they ever let us change the tool kit we'll have an example to draw from
tmacrina(TOMMY) (2025-08-16 08:14:20 UTC):
Make a video?
tmacrina(TOMMY) (2025-08-16 08:14:28 UTC):
Sounds cool
bun(TOMMY) (2025-08-18 15:40:55 UTC):
@Tommy Macrina <https://www.youtube.com/watch?v=l8dv8aKvwy0>
tmacrina(TOMMY) (2025-08-20 12:41:16 UTC):
Have you not been able to get the <https://docs.webknossos.org/webknossos/volume_annotation/tools.html#volume-interpolation|interpolation tool> to work?
tmacrina(TOMMY) (2025-08-20 12:41:54 UTC):
I like your idea, but I'd be curious if you think it would speed you up dramatically from the interpolation tool.
tmacrina(TOMMY) (2025-08-20 12:47:03 UTC):
The idea of being able to move an extruded cross section seems like a useful new feature.
tmacrina(TOMMY) (2025-08-20 12:54:15 UTC):
If the drag feature was implemented, how critical is the polygon manipulation vs the basic paint/erase functions?
bun (2025-08-20 15:06:33 UTC):
I believe it to be a critical part of the tool.

m9 — Thread in putracers

Date: 2025-08-04 18:54 UTC
View in Slack ↑ Back to Summary
bun (2025-08-04 18:54:04 UTC):
should i over segment this object in the cell body? i'm not 100% sure its a mitochondria. it might be a peroxisome or lysosome or something else :thinking_face: @Tommy Macrina @Eric
tmacrina(TOMMY) (2025-08-05 19:33:04 UTC):
Link or location?
tmacrina(TOMMY) (2025-08-05 19:34:15 UTC):
It's not jumping out at me in your volume. I think your volume looks pretty good.
bun (2025-08-05 19:34:48 UTC):
295127, 258550, 3573
bun(TOMMY) (2025-08-05 19:35:24 UTC):
i went ahead and segmented it. figured we could always get rid of it if we need to.
tmacrina(TOMMY) (2025-08-05 19:35:29 UTC):
Yeah, I think you make the right call to segment it.
bun (2025-08-05 19:36:40 UTC):
i think my volume is ready for review if that's segment is all good

m10 — Thread in putracers

Date: 2025-08-05 14:31 UTC
View in Slack ↑ Back to Summary
Jay (2025-08-05 14:31:09 UTC):
@Eric @Tommy Macrina Point of clarification: should all mitochondria be segmented with the same segment/color, or should they each be given an individual segment/color? Sorry, it's been so many years since I've done one of these tasks that I don't remember what the correct protocol is
tmacrina(TOMMY) (2025-08-05 19:27:41 UTC):
Different segment for each mitochondria.

m11 — Thread in putracers

Date: 2025-08-05 19:38 UTC
View in Slack ↑ Back to Summary
tmacrina(TOMMY) (2025-08-05 19:38:27 UTC):
Does anyone have a "hallucination" that they found particularly difficult?
bun (2025-08-05 19:52:18 UTC):
294885, 258672, 3576, 294908, 258553, 3589, 295009, 258619, 3565
bun (2025-08-05 19:52:41 UTC):
just off the top of my head. these areas were a bit tricky
Ryan (2025-08-06 15:09:27 UTC):
So far, because these slices are artificial, I've just done my best to annotate something that makes sense with the flow/movement and what makes sense with known morphology, even if it slightly goes against the shape/borders the up-sampling came up with

m12 — Thread in putracers

Date: 2025-08-05 20:03 UTC
View in Slack ↑ Back to Summary
tmacrina(TOMMY) (2025-08-05 20:03:05 UTC):
How much have you all used the ad hoc meshing in webknossos?
Jay (2025-08-05 20:14:40 UTC):
The interpolation? I tried it out for a while, but I found it produced enough errors that I was spending just as much time cleaning it up as I would be tracing the slices myself so I stopped using it
tmacrina(TOMMY) (2025-08-05 20:17:36 UTC):
No. If you right-click on a segment, you can generate a 3D mesh.
Jay (2025-08-05 20:24:52 UTC):
I didn't even realize there was a 3D view. How do you access it? I don't see any menu options, and the "keyboard shortcuts" help option is just a 404 error
tmacrina(TOMMY) (2025-08-05 20:46:19 UTC):
Hit "." to toggle between all views and the view you hover over.
tmacrina(TOMMY) (2025-08-05 20:46:28 UTC):
Docs for webknossos are online, too
tmacrina(TOMMY) (2025-08-05 20:46:29 UTC):
<https://docs.webknossos.org/webknossos/ui/keyboard_shortcuts.html>

m13 — Thread in putracers

Date: 2025-08-05 20:20 UTC
View in Slack ↑ Back to Summary
tmacrina(TOMMY) (2025-08-05 20:20:14 UTC):
Before calling these tasks finished, would you mind having at least one other team member inspect your work? You might share the places where you struggled the most.
bun(TOMMY) (2025-08-05 20:39:13 UTC):
can you remind us how to do that in webknosses. i can't seem to "View" other peoples tasks
tmacrina(TOMMY) (2025-08-05 20:45:02 UTC):
Go to tasks, search for tasks under project `mec_cutout_0_upsampled`, click on plus symbol for any task, go to `actions` button on the right side, select `View`.
bun (2025-08-05 20:48:52 UTC):
thanks!

m14 — Thread in putracers

Date: 2025-08-06 14:19 UTC
View in Slack ↑ Back to Summary
Austin.Burke (2025-08-06 14:19:29 UTC):
Ok I believe that my cube is finished and ready for review. I accidently hit finish before I was done, but was able to get back in and edit the volume with the correct annotations. my volume ID is 68838491010000ce007908ba
bun (2025-08-06 14:43:32 UTC):
I'll review
bun (2025-08-06 14:48:55 UTC):
i see the mitochondria aren't labeled is this the most up to date version?
Austin.Burke (2025-08-06 14:54:24 UTC):
<http://wk.zetta.ai:9000/links/Tu-sDfC6zgFtRV8C|68838491010000ce007908ba> thats the link to my most up to date one
Austin.Burke (2025-08-06 14:54:58 UTC):
im sure that has something to do with me clicking finish before actually being done
bun (2025-08-06 14:55:34 UTC):
ty!
bun (2025-08-06 14:59:05 UTC):
hmm your link has the 2D as black. any idea how to change it :o
Austin.Burke (2025-08-06 14:59:40 UTC):
uhhhh now thats odd
bun (2025-08-06 15:01:44 UTC):
@Tommy Macrina do you know how to make share links keep the 2D outside of the BB?
Austin.Burke (2025-08-06 15:02:08 UTC):
oh the 2D outside the BB? yea that probably got deleted when I accidently hit finish
Austin.Burke (2025-08-06 15:02:40 UTC):
is there some way to like port a set of annotation to a new dataset? cause if so I can just redo the task and move the annotations over
bun (2025-08-06 15:05:34 UTC):
hmm im not sure. what was your process for reopening the project?
bun (2025-08-06 15:06:15 UTC):
The link you sent is called Actually Finished GT. How did you name the volume lol
Austin.Burke (2025-08-06 15:07:01 UTC):
i posted to volume ID and then realised it wasnt the link and then tried to paste the link over the ID but it hyperlinked it lol
bun (2025-08-06 15:07:46 UTC):
lol
Austin.Burke (2025-08-06 15:08:13 UTC):
I reopened the annotation by going to my annotation list and then clicking it to view and that was the only way get the paint tools back
bun (2025-08-06 15:11:10 UTC):
what about the "Reopen" button under the + sign? i think that's how you start a finished task if you need to edit
Austin.Burke (2025-08-06 15:14:51 UTC):
this is the new task I reopened with zero annotation
Austin.Burke (2025-08-06 15:15:12 UTC):
and this is the "finished" task section. no reopen option
bun (2025-08-06 15:20:21 UTC):
is the only difference between the volumes the mitochondria are segmented? if so ill use the old volume to check things
Austin.Burke (2025-08-06 15:21:02 UTC):
no because I hit finish after erasing some but not all. so there area few slides but then the rest are blank
Austin.Burke (2025-08-06 15:21:04 UTC):
trust me
Austin.Burke (2025-08-06 15:21:10 UTC):
I already wanna break my laptop
Austin.Burke (2025-08-06 15:21:15 UTC):
for being this dumb
bun (2025-08-06 15:21:51 UTC):
lol it's all good. i'll just make notes and if it's already fixed no biggy
bun(TOMMY) (2025-08-06 15:22:07 UTC):
honestly it's webknosses UI that needs some fixin' clearly
tmacrina(TOMMY) (2025-08-06 15:22:36 UTC):
No idea how that link doesn't want to load the image data.
bun (2025-08-06 15:22:45 UTC):
the important buttons are hidden behind a tiny plus sign and a dropdown menu which you have to go to admin page to enable through a manual search for your task lol
tmacrina(TOMMY) (2025-08-06 15:22:57 UTC):
If you work through tasks > actions > view, it shows up: <http://wk.zetta.ai:9000/annotations/688a4519010000db0079097b#295030,259728,3539,0,0.35>
bun (2025-08-06 15:23:16 UTC):
that is an old ssegmentation
bun (2025-08-06 15:30:20 UTC):
@Tommy Macrina is there a way to get a quick3D view. I remember reviewing in 3D can help you find glaring errors really fast.
tmacrina(TOMMY) (2025-08-06 15:40:04 UTC):
Hit `.`
tmacrina(TOMMY) (2025-08-06 15:40:59 UTC):
Ohhh... maybe Austin copied the task to his own account. That could explain the missing image.
bun (2025-08-06 15:43:14 UTC):
There are no segments here:
bun (2025-08-06 15:43:38 UTC):
nm i figured it out

m15 — Thread in putracers

Date: 2025-08-06 15:17 UTC
View in Slack ↑ Back to Summary
bun (2025-08-06 15:17:12 UTC):
As for the GT best practices I know Tommy mentioned the cutouts shouldn't have had that description applied: "Segment all cells. Segment myelin as a separate object, without padding between the myelin and neighboring objects. Pay special attention to segment any invaginations, or invagination-like structures. Do NOT oversegment mitochondria. Leave extracellular space unsegmented." I think the most up to date ground truth best practice would be this: "Segment all cells. Segment myelin as a separate object, *without padding between* the myelin and neighboring objects. Pay special attention to segment any invaginations, or invagination-like structures. Segment mitochondria *without membrane painted*. Segment extracellular space." @Tommy Macrina Is that fair to say?
Ryan (2025-08-06 15:29:49 UTC):
I think we're supposed to avoid the mitochondria's border/membrane: <https://eyewire.slack.com/archives/C0A4SQRT2/p1754062100372739?thread_ts=1754061929.515139&cid=C0A4SQRT2>
Jay (2025-08-06 15:30:26 UTC):
Yeah to clarify, I think the following schema should be correct for myelin: 1. No padding between myelin and the cell it surrounds 2. No padding between myelin and other myelin 3. Myelin should butt up against the cell membranes of other cells, but not touch the actual segment
bun (2025-08-06 15:35:06 UTC):
Yes I think this is mostly correct. The outside of myelin shouldn't touch other cells/membranes, but it might be the case that it can touch other myelin. Not 100% about that though. The inside should touch the cell it is myelinating like shown. @Tommy Macrina
bun (2025-08-06 15:35:41 UTC):
@Ryan yes, "Segment mitochondria *without membrane painted*."
Ryan(TOMMY) (2025-08-06 15:36:34 UTC):
Oh yes I definitely misread that!
tmacrina(TOMMY) (2025-08-06 15:39:49 UTC):
Yeah, that's right, @bun

m16 — Thread in putracers

Date: 2025-08-14 17:58 UTC
View in Slack ↑ Back to Summary
Jay (2025-08-14 17:58:06 UTC):
@Jay @bun @Austin.Burke @teammate @Ryan This round of left-right matching is complete, so we're switching over to the <http://wk.zetta.ai:9000/projects/689da711010000a200790cad/tasks|Zetta tracing>. I'm assuming the directions are the same as before (segment mitochondria all as their own segments, no padding between myelin and the cell it ensheathes), but if not @Tommy Macrina can correct me. If another round of left-right matching comes in we'll pause this task to complete it and then move back to tracing, as per Chris' instructions.
bun (2025-08-14 17:59:05 UTC):
I can't get webknossos to load @Tommy Macrina do we need to update our IP again?
wileykit (2025-08-14 18:00:11 UTC):
Round 3 is up on the matching
Jay (2025-08-14 18:00:30 UTC):
welp lol
Jay (2025-08-14 18:00:45 UTC):
Guess we're back to matching for now then!
Ryan (2025-08-14 18:01:52 UTC):
Whew that was quick :smile:
Jay (2025-08-14 18:05:15 UTC):
Ok, spreadsheet is formatted, but I'm heading over to coding club until 3-4pm
wileykit(TOMMY) (2025-08-14 18:08:54 UTC):
ditto
tmacrina(TOMMY) (2025-08-14 21:30:22 UTC):
@bun yeah, if your IP address changes, it needs to be updated.
bun (2025-08-14 21:42:11 UTC):
~my IP hasn't changed if im looking at this correctly~
bun (2025-08-14 21:45:46 UTC):
nm it totally updated - and updating it worked

m17 — Thread in putracers

Date: 2025-08-18 19:37 UTC
View in Slack ↑ Back to Summary
Austin.Burke (2025-08-18 19:37:34 UTC):
@Tommy Macrina What would be the protocol for EM damage like this? Should it be left alone as is or segmented as the same segID or a separate one?
tmacrina(TOMMY) (2025-08-20 13:16:58 UTC):
Agree with @bun below

m18 — Thread in putracers

Date: 2025-08-31 19:07 UTC
View in Slack ↑ Back to Summary
ranl (2025-08-31 19:07:47 UTC):
<!channel>, I deployed a experimental webknossos server at <http://ranl.zetta.ai:9000/>, upgraded to the new 25.09 release and include new features like <https://forum.image.sc/t/ai-quick-select-tool-in-webknossos-now-in-3d/101290|SAM2>. The new server does not contain any old data, so you have to register again. Migrate the database might be difficult so we probably switch after the current tracing tasks are finished, meanwhile you can give it a try and report if there are any issues.
tmacrina(TOMMY) (2025-09-02 10:38:04 UTC):
@ranl Can you give me access?
ranl(TOMMY) (2025-09-02 12:49:03 UTC):
done
tmacrina(TOMMY) (2025-09-02 13:56:33 UTC):
Thanks!
tmacrina(TOMMY) (2025-09-02 14:06:09 UTC):
Pretty good at bigger convex cross sections without many intercellular objects. Cool!
tmacrina(TOMMY) (2025-09-02 14:06:29 UTC):
Got it working on some objects for 20 sections just pointing & clicking.
bun (2025-09-03 13:11:28 UTC):
i'll test it today. just registered
bun (2025-09-03 13:54:29 UTC):
this is so good
bun (2025-09-03 13:55:42 UTC):
could we also get the bounding box fill tool. from what i've been seeing it can reduce the time it takes for the AI to paint. i'm not sure exactly how it works but it limits the paint to just what's within the bound box I think.
ranl (2025-09-03 14:16:00 UTC):
You mean giving a bbox instead of point for AI, it is still there, may acting weird when you apply to fast moving object in many sections
bun (2025-09-03 14:16:40 UTC):
<https://www.youtube.com/watch?v=YByc_ERY9dg>
bun (2025-09-03 14:16:48 UTC):
like this but when using the AI tools
ranl (2025-09-03 14:23:25 UTC):
It is doable, but it will not speed up the AI, zoom in will help to reduce the data sent to ai, but the result may degrade too
ranl (2025-09-04 18:46:01 UTC):
Added a new function AI to use existing segment annotations as hints, the quick selection tool will send the segment you are current annotating to AI and AI will use it as a mask or sample additional points from it to grow the segment. Right now it only use the segment in the current section, either created by AI or manually annotated. Should help to make the AI behave consistently with myelin and mitochondria.
bun (2025-09-04 18:47:01 UTC):
ooooo! that sounds really good
bun (2025-09-04 18:47:26 UTC):
this is on the experimental server i presume?
ranl (2025-09-04 18:47:39 UTC):
yes
bun (2025-09-04 19:10:58 UTC):
im messin' around with the tools on the experimental server. and it feels like the regular quick select has some tighter traces compared to the AI quick select in it's current state. it's also really fast loading. is the center of the square still the target of the filling logic for the AI quick select toggle? the little cross hair centered in the quick select, i noticed it wasn't there with the AI quick select toggle. and for the starter segment hint - how do we know if it's using the hint or not. do i need to place a certain amount of paint/slices down to get a good anchor per say. currently i think the quick select is the fastest most accurate but requires more micro. i think if we can refine the AI's guesses to be as good as how good the current quick selecting logic is then every slice would be super accurate and clean. as a side note i saw there were some skeleton tools. it made me think maybe - since it looks like the box quick select uses a center point as it's target - would it make sense to have a kind of skeleton/box quick select hybrid tool. which would seed the center points and then does a batched quick select? idk just thinking out loud now - cool stuff so far!!!
ranl (2025-09-04 19:18:29 UTC):
You can try to increase the "Confidence Threshold" to make the AI selects less voxels. If the "Extra Condition" is set to "none", the place you click will be the seeding point for AI to grow the segment. For "points" 10 extra seeds will be sampled from the existing segment in the current view, so in general the result will be more aggressive. "mask" will use the existing segment in the current view as a mask, it is unlikely points outside of it will be included, so it is a negative constraint in general. There are some discussions about use skeletons on webknossos github page: <https://github.com/scalableminds/webknossos/issues/7541> they eventually decided it is not good enough I guess
ranl (2025-09-26 01:20:08 UTC):
I added erosion to the quick selection AI mode, also allow previewing and dynamically adjusting the confident threshold and erode parameter.
bun (2025-09-26 13:59:01 UTC):
very cool! erode decreases size (this seems like what i imagine i'll use the most). what does confidence do? i assume it sets the over all possible amount of segmentation for lack of a better word.
ranl (2025-09-26 15:19:28 UTC):
Yes, the value represent the minimal confident the SAM2 model believes the voxels under the mask belong to the object you click on

⚠️ Note on Accuracy

All message text is copied verbatim from the Slack export, including:

Document Information

Source: Slack workspace (#putracers channel)
Date Range: August 1 - September 3, 2025
Primary Subject Matter Expert: Tommy Macrina
Compilation Date: November 6, 2025
Total Messages Referenced: 18 (6 standalone + 12 threaded discussions)

Images: Referenced images stored in tommy_macrina_images/ directory