用自蒸馏和图像增强提升细粒度分割精度，40秒内看懂关键改进
SAM3 Self-Distillation for Fine-Grained GOOSE 2D Semantic Segmentation
- 以SAM3为教师模型，自蒸馏优化低效类别预测
- 测试时多尺度重缩放使固定输入模型实现多尺度推理，提升69.73% mIoU
- 引入2025冠军的强光照扭曲数据增强，是最大性能提升来源
我们提交了2026年ICRA GOOSE 2D细粒度语义分割挑战赛的第四名方案，在官方1,815张图像测试集上达到69.73%的复合平均交并比（mIoU）。模型基于近期视觉基础模型Segment Anything Model 3（SAM3）的图像编码器，搭配轻量解码器。此外，我们提出两项技术与一项经验发现：(i) 自蒸馏策略，利用在某些类别上表现更优的SAM3自身作为教师，通过真实标注框提示进行指导；(ii) 图像级多尺度测试时增强方法，通过缩放图像而非模型输入来恢复多尺度推理能力；(iii) 发现将2025年冠军方案中的强光度失真方法移植到本管线，是性能提升的最大单一来源。
原文摘要 · Abstract (English)
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.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。