TeraiNet

Use when working with camera-trap datasets, wildlife species labels, data sampling, location-aware splits, image filtering, or classifier evaluation. Preserves data integrity and prevents leakage.

Use when packaging a TeraiNet model release, building or reviewing inference code, exporting class mappings, validating model artifacts, or preparing deployment to Hugging Face or AddaxAI.

Use when converting Jupyter notebook experiments into reusable TeraiNet source code, removing hidden notebook state, extracting preprocessing/training/inference helpers, or adding tests for notebook logic.

Use when modifying TensorFlow or Keras image-classification models, tf.data pipelines, augmentations, checkpoints, experiment tracking, or evaluation. Ensures reproducible training and inference consistency.