change logger to also use csv

remove all checkpoint files from earlier runs

ignore .chpt files

added in_memory loading to reduce disk reads and increase speed

Add code so loss function uses torch.Size([x,1]) instead of torch.Size([x])

add testing notebook for convolutional neural network

run experiments comparing MSE and MAE

add experiment to compare MSE and MAE (focus150.yml and focusMSE_150.yml)

add FNN with MSE loss

add hyperparameter tuning experiment with more data for focus

change docker-compose to not use gpu

add dockerignore

add missing changes from original repository

add so far done experiments

add proper ignore for attributes and logs

add data metadata

cleanup to make ready as submodule

initial commit

add current state of metadata

rename metric to mae

fix predictions to not use argmax

add configuration files for model

implement simple model

implement data loader

Add dataset in preparation for the data module

change test_object to actual ones in file path

add .env.example files

make docker-files runnable on GPUs

add docker setup for training with tensorboard status

remove old template and use https://github.com/ashleve/lightning-hydra-template instead

add dockerfile for model

add script to convert annotations to csv

add host networking

add focus annotator as submodule to have thesis code in one place

change addressing from dataframe fix object name being incorrect

fix path for image loading being wrong

remove print from compute overlap

do not add out path to paths in json file

add overlap threshold for tile selection

change code to select based on study id instead of date

fix missing / when copying multiple files

add dockerignore file to avoid slowing down the build unnecessarily

add readme file to extraction folder

fix joining on wrong id

add variable to decide root of imag efilepaths and move export to outfolder instead of root dir

add gitignore file

add requirements

add possibility to run in container

add python scripts

add empty giignore to out folder so it does exists