Model Inference#
Adding model predictions to your dataset is the bridge between training and
evaluation. If your model is in the Model Zoo,
dataset.apply_model() handles inference in one line β including with your
own fine-tuned weights. For custom models, write a simple inference loop that
converts your modelβs outputs into FiftyOne labels.
If youβre using FiftyOne Enterprise with cloud-backed media π, download media locally within your inference loop using the media caching utilities.