Yolo_predict_tutorial_deepsea #6258

mjoudy · 2025-07-16T16:28:11Z

Hi all,

This is a tutorial for yolo prediction tool based on deepsea seanoe dataset. Please kindly give your feedback.

…al_deepsea

kostrykin

Thanks! Some comments inside.

topics/imaging/tutorials/yolo_prediction/tutorial.md

topics/imaging/images/yolo/bythog2.jpeg

topics/imaging/tutorials/yolo_prediction/tutorial.md

Co-authored-by: Leonid Kostrykin <[email protected]>

yvanlebras · 2025-07-18T09:53:01Z

Hi Mohammad. I added previously added content from the older PR (sorry, I was not seeing there was 2 identicial PR...) proposing a format more linked with GTN material with hands-on, comment tip and warning. I also propose to be co-author and to propose this tutorial in both Imaging and Ecology GTN sections (if this seems ok for you @shiltemann @bebatut @hexylena ).

Plaese, don't hesitate to modify / comment / revert !

topics/imaging/tutorials/yolo_prediction/tutorial.md

kostrykin · 2025-07-18T11:05:59Z

topics/imaging/tutorials/yolo_prediction/tutorial.md

+> >    - `Image size`: Use 1000 (or a smaller number like 640 if processing speed is important). This controls how much the image is resized before prediction. Smaller values = faster but possibly less accurate.
+> >    - `Confidence threshold`: Set to 0.25 (25%). This controls how confident the model must be to report a detection. If you increase this value (e.g., 0.5), you’ll get fewer detections, but they’ll be more confident. If you lower it (e.g., 0.1), you may get more results, but possibly more false positives.
+> >    - `IoU threshold`: Set to 0.45. This is used for Non-Maximum Suppression (NMS), which removes overlapping detections. A higher IoU value (e.g., 0.7) keeps more overlapping boxes. A lower IoU (e.g., 0.3) removes more overlaps, which may help clean up crowded images.
+> >    - `Max detections`: Set a reasonable cap like 300. This limits the number of objects detected per image.


Please consider adapting the formatting of the parameters to the format used in lines 113–118.

on the image size, if we make prediction on an image with 512x512, setting image size to 1000 will probably not work? im not too sure..better to test it

also, i think yolo automatically adjust the size, if input 1000, yolo will change it to 1024, 500 will become 512.
maybe we need to add this to the image size explaination? we need to test it, to be safe.. i remember i had some issue with setting size bigger than the image..

Good point @sunyi000 I think its correct. I will test it and add this point to the tutorial.

@sunyi000 I tested this issue and there is no problem. my images size are 1920by1080 and in the tool I tested for 640, 1000, 2000, 3000. They all had correct outputs.

Gemfile.lock

Co-authored-by: Leonid Kostrykin <[email protected]>

anuprulez · 2025-07-18T12:25:03Z

@yvanlebras I agree. We can have this tutorial for both - Ecology and BioImaging. Actually, i was talking to @sunyi000 two days back and he suggested the same. He plans to work on a new tutorial for Yolo training tool. @mjoudy can you use images from BioImage model zoo (https://bioimage.io/#/models) to make prediction using Yolo models? Some sample images you can find in this tutorial: https://training.galaxyproject.org/training-material/topics/imaging/tutorials/process-image-bioimageio/tutorial.html

anuprulez · 2025-07-18T12:25:55Z

@yvanlebras @kostrykin thank you very much for your work on this. @mjoudy please address the issues mentioned

yvanlebras · 2025-07-22T10:07:04Z

For SEANOE data part, I can also add some steps further to take all txt files to produce a summary of number of each species detected if relevant / of interest.

mjoudy · 2025-07-23T15:06:03Z

Hi Mohammad. I added previously added content from the older PR (sorry, I was not seeing there was 2 identicial PR...) proposing a format more linked with GTN material with hands-on, comment tip and warning. I also propose to be co-author and to propose this tutorial in both Imaging and Ecology GTN sections (if this seems ok for you @shiltemann @bebatut @hexylena ).

Plaese, don't hesitate to modify / comment / revert !

Thanks Yvan. sorry I had made a mistake and opened two PR. after a while I closed one of them.
Thank you for your suggestions and help.

mjoudy · 2025-07-23T15:18:08Z

@yvanlebras I agree. We can have this tutorial for both - Ecology and BioImaging. Actually, i was talking to @sunyi000 two days back and he suggested the same. He plans to work on a new tutorial for Yolo training tool. @mjoudy can you use images from BioImage model zoo (https://bioimage.io/#/models) to make prediction using Yolo models? Some sample images you can find in this tutorial: https://training.galaxyproject.org/training-material/topics/imaging/tutorials/process-image-bioimageio/tutorial.html

Thank you Anup for suggestions. Actually, as far as I know, images and models in the BioImage zoo are not yolo-based. Please let me know if there are any. Currently, I am looking for some bio-inspired yolo pre-trained models to make a better example for the segmentation part.

Co-authored-by: Anup Kumar, PhD <[email protected]>

mjoudy · 2025-07-31T09:03:09Z

I have added a few comments. I am trying to use this tutorial and these are my observations.

We should make the datasets access easier. For the tutorial, if we include a few images, that would be enough to show the usage of the tool. Currently we are pulling a large set of around 4000 images. We can point users to the original large repository. I think these sample images and trained model files should also come from Zenodo. In the current form, it is not easy to find the used datasets. Please me know your opinions @yvanlebras @kostrykin @mjoudy . Thanks

Thanks for the suggestion. I will make a sample and provide its zenodo link.

anuprulez · 2025-08-11T08:02:47Z

Hi @yvanlebras, can we update the input datasets to include just a few images and take them from Zenodo? Then we will have the opportunity to simplify the history operations a lot that we have in the tutorial? Thanks a lot!

topics/imaging/tutorials/yolo_prediction/tutorial.md

kostrykin

Regarding the linting issue…

topics/imaging/tutorials/yolo_prediction/tutorial.md

kostrykin

I have gathered a list of open issues:

https://github.com/galaxyproject/training-material/pull/6258/files#r2230652505
https://github.com/galaxyproject/training-material/pull/6258/files#r2230658220
https://github.com/galaxyproject/training-material/pull/6258/files#r2270065822
https://github.com/galaxyproject/training-material/pull/6258/files#r2244759562
I was actually going to work on this one, but I got blocked by the one below:
https://github.com/galaxyproject/training-material/pull/6258/files#r2270096781

topics/imaging/tutorials/yolo_prediction/tutorial.md

Co-authored-by: Leonid Kostrykin <[email protected]>

yvanlebras · 2025-10-06T20:33:49Z

Hi everyone. Really sorry to come back just now on this PR! I created a new history to try on "just" 2 input images https://usegalaxy.eu/u/ylebras/h/yolo-training-on-just-2-images so we can put these files in Zenodo for a more easier to follow and use tutorial! Will try to create the GTN Zenodo repo, and / or Arthur will continue and finish this tutorial!
When I was tested, I saw that the last version of the tool, v4, is not workinf properly apparently (see error datasert on previously mentionned history)...

shiltemann

Thanks @mjoudy! I just pushed some changes to address the linting errors.

Thanks @yvanlebras and @kostrykin for reviewing testing and improving this

shiltemann · 2025-11-06T15:43:48Z

CONTRIBUTORS.yaml

    affiliations:
      - deKCD

+mjoudy:


@mjoudy I added you to the CONTRIBUTORS.yaml file, feel free to add more info here about yourself (affiliation, socials, orcid, etc)

shiltemann · 2025-11-06T15:44:39Z

topics/imaging/tutorials/yolo_prediction/tutorial.md

+  - type: "internal"
+    topic_name: machine-learning
+    tutorials:
+      - ml-advanced-image


Which tutorial are you trying to refer to here? this should match the folder name the tutorial is in (same for the topic)

github-actions · 2025-11-06T16:02:34Z

topics/imaging/tutorials/yolo_prediction/tutorial.md

+
+YOLOv8 in segment mode produces a more detailed output than detection:
+
+**Segmented overlay (*.jpg):**


⚠️ [GTN Lint] <GTN:020> _{reported by reviewdog 🐶}
This looks like a heading, but isn't. Please use proper semantic headings where possible. You should check the heading level of this suggestion, rather than accepting the change as-is.

Suggested change

**Segmented overlay (*.jpg):**

**### Segmented overlay (*.jpg):**

github-actions · 2025-11-06T16:02:34Z

topics/imaging/tutorials/yolo_prediction/tutorial.md

+<img src="../../images/yolo/bus.jpg" style="width:40%;" alt="Input image of a bus used in segmentation example">
+
+
+**Mask file (*_mask.tiff):**


⚠️ [GTN Lint] <GTN:020> _{reported by reviewdog 🐶}
This looks like a heading, but isn't. Please use proper semantic headings where possible. You should check the heading level of this suggestion, rather than accepting the change as-is.

Suggested change

**Mask file (*_mask.tiff):**

**### Mask file (*_mask.tiff):**

github-actions · 2025-11-06T16:02:34Z

topics/imaging/tutorials/yolo_prediction/tutorial.md

+
+<img src="../../images/yolo/bus_mask.png" style="width:40%;" alt="YOLOv8 predicted mask on the bus image">
+
+**Annotation file (*.txt):**


⚠️ [GTN Lint] <GTN:020> _{reported by reviewdog 🐶}
This looks like a heading, but isn't. Please use proper semantic headings where possible. You should check the heading level of this suggestion, rather than accepting the change as-is.

Suggested change

**Annotation file (*.txt):**

**### Annotation file (*.txt):**

kostrykin · 2025-11-24T09:00:00Z

Hi everyone. Really sorry to come back just now on this PR! I created a new history to try on "just" 2 input images https://usegalaxy.eu/u/ylebras/h/yolo-training-on-just-2-images so we can put these files in Zenodo for a more easier to follow and use tutorial! Will try to create the GTN Zenodo repo, and / or Arthur will continue and finish this tutorial! When I was tested, I saw that the last version of the tool, v4, is not workinf properly apparently (see error datasert on previously mentionned history)...

The error is with the toolshed.g2.bx.psu.edu/repos/bgruening/yolo_predict/yolo_predict/8.3.0+galaxy4 tool, @sunyi000 can you maybe help with this?

sunyi000 · 2025-11-24T10:47:43Z

Hi everyone. Really sorry to come back just now on this PR! I created a new history to try on "just" 2 input images https://usegalaxy.eu/u/ylebras/h/yolo-training-on-just-2-images so we can put these files in Zenodo for a more easier to follow and use tutorial! Will try to create the GTN Zenodo repo, and / or Arthur will continue and finish this tutorial! When I was tested, I saw that the last version of the tool, v4, is not workinf properly apparently (see error datasert on previously mentionned history)...

The error is with the toolshed.g2.bx.psu.edu/repos/bgruening/yolo_predict/yolo_predict/8.3.0+galaxy4 tool, @sunyi000 can you maybe help with this?

from the output in history, looks like a detection model is used for segmentation tasks, so the masks are None?

mjoudy added 2 commits July 16, 2025 16:09

add yolo prediction tutorial based on deepsea dataset.

01b2f0e

Gemfile?!

37d1a8b

github-actions bot added template-and-tools imaging labels Jul 16, 2025

Merge remote-tracking branch 'upstream/main' into yolo_predict_tutori…

c40eabb

…al_deepsea

mjoudy marked this pull request as ready for review July 16, 2025 19:42

mjoudy requested review from beatrizserrano and kostrykin as code owners July 16, 2025 19:42

anuprulez self-requested a review July 18, 2025 08:05

kostrykin reviewed Jul 18, 2025

View reviewed changes

yvanlebras and others added 7 commits July 18, 2025 11:35

Update topics/imaging/tutorials/yolo_prediction/tutorial.md

868b9dc

Co-authored-by: Leonid Kostrykin <[email protected]>

Update topics/imaging/tutorials/yolo_prediction/tutorial.md

a95ce4d

Co-authored-by: Leonid Kostrykin <[email protected]>

Update topics/imaging/tutorials/yolo_prediction/tutorial.md

3547180

Co-authored-by: Leonid Kostrykin <[email protected]>

Update topics/imaging/tutorials/yolo_prediction/tutorial.md

498fe12

Co-authored-by: Leonid Kostrykin <[email protected]>

Update topics/imaging/tutorials/yolo_prediction/tutorial.md

c9083d0

Co-authored-by: Leonid Kostrykin <[email protected]>

Update topics/imaging/tutorials/yolo_prediction/tutorial.md

95c963b

Co-authored-by: Leonid Kostrykin <[email protected]>

Update tutorial.md with GTN compliant elements (I hope ;) )

52c1afa

kostrykin reviewed Jul 18, 2025

View reviewed changes

kostrykin requested changes Jul 18, 2025

View reviewed changes

Gemfile.lock Show resolved Hide resolved

yvanlebras and others added 3 commits July 18, 2025 14:09

Update topics/imaging/tutorials/yolo_prediction/tutorial.md

f31775f

Co-authored-by: Leonid Kostrykin <[email protected]>

Update topics/imaging/tutorials/yolo_prediction/tutorial.md

1b574c7

Co-authored-by: Leonid Kostrykin <[email protected]>

Update tutorial.md to homogeneize parameters presentation

29cd75c

yvanlebras added 2 commits July 18, 2025 22:54

fix some typo and add a step to create class name file

82bd6f6

fix typo and add SEANOE dataset reference

7654df9

Update topics/imaging/tutorials/yolo_prediction/tutorial.md

931ca71

Co-authored-by: Anup Kumar, PhD <[email protected]>

Merge branch 'main' into yolo_predict_tutorial_deepsea

bc8252d

galaxyproject deleted a comment from github-actions bot Aug 11, 2025

Update tutorial.md

b0f73b0

github-actions bot reviewed Aug 11, 2025

View reviewed changes

topics/imaging/tutorials/yolo_prediction/tutorial.md Outdated Show resolved Hide resolved

kostrykin reviewed Aug 11, 2025

View reviewed changes

kostrykin added 5 commits August 11, 2025 15:09

Apply suggestion from @kostrykin

6d0ef7a

Update topics/imaging/tutorials/yolo_prediction/tutorial.md

3badbb6

Update topics/imaging/tutorials/yolo_prediction/tutorial.md

d7efb00

Update topics/imaging/tutorials/yolo_prediction/tutorial.md

b2b838e

Update topics/imaging/tutorials/yolo_prediction/tutorial.md

de8b832

kostrykin reviewed Aug 12, 2025

View reviewed changes

topics/imaging/tutorials/yolo_prediction/tutorial.md Outdated Show resolved Hide resolved

topics/imaging/tutorials/yolo_prediction/tutorial.md Outdated Show resolved Hide resolved

kostrykin reviewed Aug 14, 2025

View reviewed changes

topics/imaging/tutorials/yolo_prediction/tutorial.md Outdated Show resolved Hide resolved

topics/imaging/tutorials/yolo_prediction/tutorial.md Outdated Show resolved Hide resolved

topics/imaging/tutorials/yolo_prediction/tutorial.md Outdated Show resolved Hide resolved

kostrykin and others added 4 commits August 14, 2025 16:37

Update topics/imaging/tutorials/yolo_prediction/tutorial.md

0ad22ca

Apply suggestions from code review

596b41b

Update topics/imaging/tutorials/yolo_prediction/tutorial.md

8b8d8d0

Co-authored-by: Leonid Kostrykin <[email protected]>

Merge branch 'main' into yolo_predict_tutorial_deepsea

d406090

shiltemann added 2 commits November 6, 2025 16:36

formatting tweaks

21b0186

add mjoudy as contributor

a0d343c

github-actions bot added the template-and-tools label Nov 6, 2025

shiltemann added 2 commits November 6, 2025 16:40

typo

10eb70d

invalide metadata

d6cadcf

shiltemann reviewed Nov 6, 2025

View reviewed changes

formatting tweaks

a3a8957

github-actions bot reviewed Nov 6, 2025

View reviewed changes


		YOLOv8 in segment mode produces a more detailed output than detection:

		*Segmented overlay (.jpg):**

	*Segmented overlay (.jpg):**
	*### Segmented overlay (.jpg):**

		<img src="../../images/yolo/bus.jpg" style="width:40%;" alt="Input image of a bus used in segmentation example">


		*Mask file (_mask.tiff):**

	*Mask file (_mask.tiff):**
	*### Mask file (_mask.tiff):**


		<img src="../../images/yolo/bus_mask.png" style="width:40%;" alt="YOLOv8 predicted mask on the bus image">

		*Annotation file (.txt):**

	*Annotation file (.txt):**
	*### Annotation file (.txt):**

Yolo_predict_tutorial_deepsea #6258

Are you sure you want to change the base?

Yolo_predict_tutorial_deepsea #6258

Uh oh!

Conversation

mjoudy commented Jul 16, 2025

Uh oh!

kostrykin left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

yvanlebras commented Jul 18, 2025

Uh oh!

Uh oh!

kostrykin Jul 18, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

sunyi000 Jul 18, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

sunyi000 Jul 18, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

mjoudy Jul 23, 2025

Choose a reason for hiding this comment

Uh oh!

mjoudy Jul 24, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

anuprulez commented Jul 18, 2025

Uh oh!

anuprulez commented Jul 18, 2025

Uh oh!

yvanlebras commented Jul 22, 2025

Uh oh!

mjoudy commented Jul 23, 2025

Uh oh!

mjoudy commented Jul 23, 2025

Uh oh!

mjoudy commented Jul 31, 2025

Uh oh!

anuprulez commented Aug 11, 2025

Uh oh!

Uh oh!

kostrykin left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

kostrykin left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

yvanlebras commented Oct 6, 2025

Uh oh!

shiltemann left a comment

Choose a reason for hiding this comment

Uh oh!

shiltemann Nov 6, 2025

Choose a reason for hiding this comment

kostrykin Jul 18, 2025 •

edited

Loading

sunyi000 Jul 18, 2025 •

edited

Loading

sunyi000 Jul 18, 2025 •

edited

Loading

kostrykin left a comment •

edited

Loading