mlsanityData to 14 Oct 2026

SAM3 Self-Distillation for Fine-Grained GOOSE 2D Semantic Segmentation

arXiv 2606.20130 ↗First posted 18 Jun 2026cs.CVcited by 0

On the map: Knowledge distillation → Detection and dense prediction distillation

Abstract (arXiv)

We describe our 4th-place entry to the ICRA 2026 GOOSE 2D Fine-Grained Semantic Segmentation Challenge, which reached a composite mean Intersection-over-Union (mIoU) of 69.73% on the official 1,815-image test set. Our model adapts the image encoder of a recent visual foundation model, Segment Anything Model 3 (SAM3), with a lightweight decoder. Beyond this, we contribute two techniques and one empirical finding: (i) a self-distillation scheme that re-uses SAM3 itself, prompted with ground-truth boxes, as a teacher on the classes where it outperforms our own model; (ii) an image-level multi-scale test-time augmentation scheme that restores multi-scale inference for a fixed-input-size model by rescaling the image rather than the model input; and (iii) the finding that an aggressive photometric distortion from a winning 2025 GOOSE 2D entry, transplanted onto our pipeline, is its single largest source of improvement.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Detection and dense prediction distillation, one of 12 directions in Knowledge distillation. Builds on 13 of its 14 references that are in the map. First posted on arXiv 18 Jun 2026.

Builds on

13 of 14 references.

References · 13

  1. 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,974
  2. 2016Feature Pyramid Networks for Object Detectioncited by 5,036
  3. 2015SSD: Single Shot MultiBox Detectorcited by 3,130
  4. 2021SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformerscited by 2,940
  5. 2021Masked-attention Mask Transformer for Universal Image Segmentationcited by 2,351
  6. 2018Unified Perceptual Parsing for Scene Understandingcited by 1,271
  7. 2023Self-Supervised Learning from Images with a Joint-Embedding Predictive Architecturecited by 1,039
  8. 2025SAM 3: Segment Anything with Conceptscited by 906
  9. 2022InternImage: Exploring Large-Scale Vision Foundation Models with Deformable Convolutionscited by 482
  10. 2023The GOOSE Dataset for Perception in Unstructured Environmentscited by 37
  11. 2023Balancing Logit Variation for Long-tailed Semantic Segmentationcited by 26
  12. 2024Excavating in the Wild: The GOOSE-Ex Dataset for Semantic Segmentationcited by 13
1 more
  1. 2025Technical Report for ICRA 2025 GOOSE 2D Semantic Segmentation Challenge: Boosting Off-Road Segmentation via Photometric Distortion and Exponential Moving Averagecited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2024TTT-KD: Test-Time Training for 3D Semantic Segmentation through Knowledge Distillation from Foundation Models
  2. 2026TAG: Calibrating SAM3 for Universal Image Segmentation without Dense Supervision
  3. 2025The SAM2-to-SAM3 Gap in the Segment Anything Model Family: Why Prompt-Based Expertise Fails in Concept-Driven Image Segmentation
  4. 2025Fast SAM2 with Text-Driven Token Pruning
  5. 2026Heterogeneous and Adept Snapshot Distillation for 3D Semantic Segmentation
  6. 2024SAM-Guided Masked Token Prediction for 3D Scene Understanding
  7. 2023Knowledge Distillation from 3D to Bird's-Eye-View for LiDAR Semantic Segmentation
  8. 2023TinySAM: Pushing the Envelope for Efficient Segment Anything Model