Downloads · 30 days
4
12% of all-time downloads
morpeN1/autolens-dinov3-safe-focal
autolens-dinov3-safe-focal is a image classification model from morpeN1. Use it when you need a label for an image. It is set up for onnxruntime. The card lists the license as other.
Pre-calibration HF model card for the focal-loss DINOv3 candidate.
Downloads · 30 days
4
12% of all-time downloads
All-time downloads
34
Public
Parameters
21.6M
173 MB on disk
Likes
0
Public
Click a slice to open those files.
.onnx86.6 MB · 50%
From the Hugging Face model README
Pre-calibration HF model card for the focal-loss DINOv3 candidate.
vit_small_patch16_dinov3)dinov3_safe_focal_aug_20260510_111158checkpoints/dinov3_safe_focal_aug_20260510_111158/best-32-0.9537.ckptmodel.safetensors plus metadata.json, with ONNX for deployment[0.4429, 0.4354, 0.437][0.2456, 0.2421, 0.2449]Validation was used to select the checkpoint. The metrics below come from the best validation epoch in metrics.csv.
Evaluated on the held-out internal test set (3170 samples, no data leakage).
| Class | Precision | Recall | F1-score | Support |
|---|---|---|---|---|
| SUV | 0.9171 | 0.9582 | 0.9372 | 670 |
| VAN | 1.0000 | 0.9670 | 0.9832 | 455 |
| STATION WAGON | 0.8133 | 0.9242 | 0.8652 | 66 |
| MICRO | 0.9524 | 0.9091 | 0.9302 | 22 |
| OPEN WHEEL / F1 | 0.9983 | 0.9949 | 0.9966 | 586 |
| SEDAN | 0.9423 | 0.9761 | 0.9589 | 836 |
| HATCHBACK | 0.9156 | 0.8158 | 0.8628 | 266 |
| PICK UP | 0.9474 | 0.8699 | 0.9070 | 269 |



model.safetensors: 82.373 MBmodel.onnx: 82.572 MBmodel.safetensors — weights-only model artifactmetadata.json — architecture, classes, preprocessing, source run, and artifact metadatamodel.onnx — ONNX Runtime inference artifactmetrics.csv — epoch-level training and validation historyper_class_metrics.txt — internal-test classification reporttraining_loss.png — training/validation loss curvetraining_accuracy.png — training/validation accuracy curveconfusion_matrix.png — normalized internal-test confusion matrixsize_check.json / size_report.json — artifact size and ONNX smoke evidenceinternal_test_results.json — structured internal-test summaryEducational demo and report evidence for classifying uploaded vehicle images into the 8 AutoLens body-type classes.
The dataset is assembled from public/open sources and may contain domain bias. Similar body styles such as hatchback, station wagon, and sedan can be ambiguous. This repo is intentionally pre-calibration; probability calibration will be added in the next upload after temperature scaling is fitted.