Feature/SK-971 | New object detection example #703
Conversation
|
Left to do:
|
|
|
||
| ### 1. Prerequisites | ||
|
|
||
| - `Python >=3.8, <=3.12 <https://www.python.org/downloads>`__ |
There was a problem hiding this comment.
Have it been tested with these versions?
|
|
||
| ### 4. Running the project on FEDn | ||
|
|
||
| To learn how to set up your FEDn Studio project and connect clients, take the quickstart tutorial: https://fedn.readthedocs.io/en/stable/quickstart.html. When activating the first client, you will be asked to provide your login credentials to Kaggle to download the welding defect dataset and split it into separate client folders. |
There was a problem hiding this comment.
I would upload the dataset to scaleout-public bucket instead, so that we are in control of the data. But this is good for now.
|
|
||
| ## Experiments with results | ||
|
|
||
| Below are a few examples of experiments which have been run using this example. A centralized setup has been used as baseline to compare the federated setup to. Two clients have been used in the federated setup and a few different epoch-to-round ratios have been tested. |
There was a problem hiding this comment.
"A centralized setup has been used as baseline to compare against."
| - pip | ||
| - setuptools | ||
| - wheel | ||
| dependencies: |
There was a problem hiding this comment.
look at other examples from main branch, use the same environment markers.
| - wheel | ||
| dependencies: | ||
| - torch==2.3.1 | ||
| - torchvision==0.18.1 |
There was a problem hiding this comment.
why torchvision? Where is it used?
|
I think that we should consider having a separate repository for this example, since it uses a library (Ultralytics) with an AGPL-license (can make some users nervous, even if it is not used in FEDn itself). |
Description
New example to stack. Defect detection use case for manufacturing using YOLOv8n model.
Link to JIRA --> https://scaleoutsystems.atlassian.net/browse/SK-971?atlOrigin=eyJpIjoiMzQ0NWJhNDA0NTU2NGUyY2E2MzliNzBlZmUyZTk0NTUiLCJwIjoiaiJ9