Workflow Options

Choose between training a new model from scratch or using a previously trained model.

You have two options for how to proceed with segmentation

Train from Scratch: Choose this option when you have training data (images with corresponding annotation masks) and want to train a new model specifically for your data. This is the best choice when your images have unique characteristics that differ from standard datasets, or when you need optimal accuracy for your specific use case. Training requires labeled examples but produces a model tailored to your data.

Use Pretrained Model: Choose this option when you already have a trained model (.pth file) and its configuration (.json file) from a previous training session. This allows you to skip the training process entirely and go directly to running inference. This is useful when you want to apply a previously trained model to new data, or when sharing models between team members.