| tags: | |
| - autogluon | |
| - image | |
| - automl | |
| library_name: autogluon | |
| license: apache-2.0 | |
| pipeline_tag: image-classification | |
| datasets: | |
| - cassieli226/2025-24679-image-dataset | |
| metrics: | |
| - accuracy | |
| - f1 | |
| # HW2 — Image AutoML | |
| ## Description | |
| Trained with **AutoGluon** on an augmented image split (train/test) and validated on the original split. Backbone: `timm/resnet18`. | |
| - **Problem Type**: `image-classification` | |
| - **Preset**: `medium_quality` | |
| ## Data | |
| - Train samples: 256 | |
| - Test samples: 64 | |
| - External (original) samples: 40 | |
| - Dataset: `cassieli226/2025-24679-image-dataset` | |
| ## Results | |
| - **Test (accuracy / f1_weighted)**: 1.0000 / 1.0000 | |
| - **Original (accuracy / f1_weighted)**: 1.0000 / 1.0000 | |