Notebooks to upload/download marine footage, connect to a citizen science project, train machine learning models and publish marine biological observations.
This merge request contains all the commits to create a working environment locally, to be able to run all the notebooks. This resolves Issues: #191 and number 49 from kso-utils.
The requirement.txt files in the separate kso-object-detection and kso-data-management repos are deleted. A gpu_requirements_user.txt is added so that the user can specify the pytorch version that is needed if they have a gpu.
Instructions are added to the ReadMe so that a user can set up this working environment. Note that some specifics need to be added for Linux and Mac (if this is needed, I can't test that). See issue #194.
The first 2/3 cells of all notebooks are reorganized and changed so that it also sets up the correct environment in google Colab, and that it does not run any code 2x. The logic behind this is in the commit messages.
These 2/3 cells are now the same in all notebooks, and they clone the kso-object-detection repo when you are working on Colab (we will merge the two repos next week anyway).
Some utils functions are updated to resolve some small bugs that I ran into when testing the notebooks. these bugs are described in the commit messages.
All notebooks 1-8 are tested locally with this environment and they work. (except for some bugs that I think to be unrelated to package versions). This was tested on a windows computer (windows 11) with a GPU.
Notebooks 1-3 were also tested in Colab.
It is not tested yet on SNIC, since we need to change the docker file for this. See new issue #193.
This merge request contains all the commits to create a working environment locally, to be able to run all the notebooks. This resolves Issues: #191 and number 49 from kso-utils.
The requirement.txt files in the separate kso-object-detection and kso-data-management repos are deleted. A gpu_requirements_user.txt is added so that the user can specify the pytorch version that is needed if they have a gpu.
Instructions are added to the ReadMe so that a user can set up this working environment. Note that some specifics need to be added for Linux and Mac (if this is needed, I can't test that). See issue #194.
The first 2/3 cells of all notebooks are reorganized and changed so that it also sets up the correct environment in google Colab, and that it does not run any code 2x. The logic behind this is in the commit messages. These 2/3 cells are now the same in all notebooks, and they clone the kso-object-detection repo when you are working on Colab (we will merge the two repos next week anyway).
Some utils functions are updated to resolve some small bugs that I ran into when testing the notebooks. these bugs are described in the commit messages.
All notebooks 1-8 are tested locally with this environment and they work. (except for some bugs that I think to be unrelated to package versions). This was tested on a windows computer (windows 11) with a GPU. Notebooks 1-3 were also tested in Colab.
It is not tested yet on SNIC, since we need to change the docker file for this. See new issue #193.