User Interface

The main parts of the AmalthAI user interface are described below.

1. Dataset Creation and Annotation

AmalthAI lets users create and preview datasets for training. For tasks that require annotations, users can annotate images with the integrated CVAT tool and save them in the project structure.

2. Model Training

Users can select a dataset and model, then start training from the interface. They can use the preselected training parameters for a simple setup or expand the advanced options to set their own values. The default parameters reflect established computer vision practices and research.

3. Model Evaluation

After training, the platform selects the best model from the completed trials and epochs using the evaluation metric. The interface displays the score so users can assess the model's performance.

4. Model Inference

After training and evaluation, users can run the model on new images from the interface. The results appear on the same page, where users can review the predictions.