Training-Free Fine-Grained Semantic Segmentations in Low Data Regimes: A FungiTastic Baseline
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914588177465344 |
|---|---|
| author | Cavada, Sebastian Pelosin, Francesco Faggi, Lapo |
| author_facet | Cavada, Sebastian Pelosin, Francesco Faggi, Lapo |
| contents | Fine-grained semantic segmentation requires both precise localization and discrimination between visually similar classes. In FungiTastic, this problem is further complicated by a long-tailed distribution and strong variation in image acquisition conditions. We propose a training-free two-stage framework that decouples segmentation from classification. SAM3 first produces class-agnostic mushroom masks using macro-taxonomic prompts, and DINOv3 then assigns fine-grained labels through prototype matching in the embedding space. To improve this stage, we apply a simple transformation of the DINOv3 feature space that improves prototype-based classification. Compared with class-specific prompting, our approach is more scalable and keeps the segmentation cost low. We report results from one-shot to few-hundred-shot regimes, providing, to the best of our knowledge, the first baseline for fine-grained semantic segmentation in low-data settings. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_22492 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Training-Free Fine-Grained Semantic Segmentations in Low Data Regimes: A FungiTastic Baseline Cavada, Sebastian Pelosin, Francesco Faggi, Lapo Computer Vision and Pattern Recognition Fine-grained semantic segmentation requires both precise localization and discrimination between visually similar classes. In FungiTastic, this problem is further complicated by a long-tailed distribution and strong variation in image acquisition conditions. We propose a training-free two-stage framework that decouples segmentation from classification. SAM3 first produces class-agnostic mushroom masks using macro-taxonomic prompts, and DINOv3 then assigns fine-grained labels through prototype matching in the embedding space. To improve this stage, we apply a simple transformation of the DINOv3 feature space that improves prototype-based classification. Compared with class-specific prompting, our approach is more scalable and keeps the segmentation cost low. We report results from one-shot to few-hundred-shot regimes, providing, to the best of our knowledge, the first baseline for fine-grained semantic segmentation in low-data settings. |
| title | Training-Free Fine-Grained Semantic Segmentations in Low Data Regimes: A FungiTastic Baseline |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2605.22492 |