Add Exception Handling for DataLoader

[Erfandarzi]

The current implementation lacks exception handling in critical areas like the DataLoader setup and the training/validation loops in the DetectionModelTrainer and DetectionModelValidator classes.

Could you please add try-catch blocks around DataLoader instantiation to catch and log any initialization errors?
Shoud be also good to include error-handling within the training and validation loops to capture and log any exceptions that occur during the model training and validation processes.

[wittenator]

Heya @Erfandarzi ,

awesome that somebody considers this interesting enough to have a closer look! I mainly use this currently for my personal research (for which I have a deadline coming up). After that I will certainly have a look at some QoL things such as better logging. If you have some ideas, you are are of course invited to submit a pull request. I currently use this library in conjecture with Guild.AI which logs error messages and warnings automatically. The Detection* classes are currently very much work in progress as well btw. If you want to try this out, I would currently probably advise on only using the Classification* classes.