[P] Best image classifier architecture right now

I want to create an image classifier which classifies the season in a regular outside image - winter, spring, summer, fall/autumn.

I’ll likely go about this by finetuning an existing model using FastAI. However, it’s super hard to understand which architecture to use.

How am I supposed to pick my approach? Does anyone have a recommendation for this task?

architecture classifier fastai finetuning image machinelearning spring