Efficient Fine-Tuning of DINOv3 Pretrained on Natural Images for Atypical Mitotic Figure Classification (MIDOG 2025 Task 2 Winner)
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866911209793519616 |
|---|---|
| author | Balezo, Guillaume Feki, Hana Bourgade, Raphaël Monnier, Lily Blons, Matthieu Blondel, Alice Decencière, Etienne Planas, Albert Pla Walter, Thomas |
| author_facet | Balezo, Guillaume Feki, Hana Bourgade, Raphaël Monnier, Lily Blons, Matthieu Blondel, Alice Decencière, Etienne Planas, Albert Pla Walter, Thomas |
| contents | Atypical mitotic figures (AMFs) represent abnormal cell division associated with poor prognosis. Yet their detection remains difficult due to low prevalence, subtle morphology, and inter-observer variability. The MIDOG 2025 challenge introduces a benchmark for AMF classification across multiple domains. In this work, we fine-tuned the recently published DINOv3-H+ vision transformer, pretrained on natural images, using low-rank adaptation (LoRA), training only ~1.3M parameters in combination with extensive augmentation and a domain-weighted Focal Loss to handle domain heterogeneity. Despite the domain gap, our fine-tuned DINOv3 transfers effectively to histopathology, reaching first place on the final test set. These results highlight the advantages of DINOv3 pretraining and underline the efficiency and robustness of our fine-tuning strategy, yielding state-of-the-art results for the atypical mitosis classification challenge in MIDOG 2025. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2508_21041 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Efficient Fine-Tuning of DINOv3 Pretrained on Natural Images for Atypical Mitotic Figure Classification (MIDOG 2025 Task 2 Winner) Balezo, Guillaume Feki, Hana Bourgade, Raphaël Monnier, Lily Blons, Matthieu Blondel, Alice Decencière, Etienne Planas, Albert Pla Walter, Thomas Image and Video Processing Computer Vision and Pattern Recognition Atypical mitotic figures (AMFs) represent abnormal cell division associated with poor prognosis. Yet their detection remains difficult due to low prevalence, subtle morphology, and inter-observer variability. The MIDOG 2025 challenge introduces a benchmark for AMF classification across multiple domains. In this work, we fine-tuned the recently published DINOv3-H+ vision transformer, pretrained on natural images, using low-rank adaptation (LoRA), training only ~1.3M parameters in combination with extensive augmentation and a domain-weighted Focal Loss to handle domain heterogeneity. Despite the domain gap, our fine-tuned DINOv3 transfers effectively to histopathology, reaching first place on the final test set. These results highlight the advantages of DINOv3 pretraining and underline the efficiency and robustness of our fine-tuning strategy, yielding state-of-the-art results for the atypical mitosis classification challenge in MIDOG 2025. |
| title | Efficient Fine-Tuning of DINOv3 Pretrained on Natural Images for Atypical Mitotic Figure Classification (MIDOG 2025 Task 2 Winner) |
| topic | Image and Video Processing Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2508.21041 |