GeoDistill-Refine: Silhouette-First Geometry Distillation for Annotation-Free Spacecraft Segmentation
2026-08-07 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionArtificial Intelligence
AI summaryⓘ
The authors developed GeoDistill-Refine, a method to improve spacecraft image segmentation without needing manual labels. They combine multiple text prompts to get consistent object masks from a larger model and then train a smaller model using shape details from these masks. Their approach reduces errors by ignoring unreliable mask information and improves accuracy in identifying object boundaries compared to simpler methods. The small trained model runs quickly, making it practical for real use.
foundation segmentation modelspseudo-masktextual promptdistillationsigned-distance-fieldskeletonizationImage IoUBoundary F1TinyUNetSpacecraft imagery
Authors
Yonglong Zhang, Zongwu Xie, Yang Liu
Abstract
Foundation segmentation models can provide supervision for spacecraft imagery without manual training masks, but their predictions vary with textual prompts and may contain geometric errors that are amplified during distillation. This paper presents GeoDistill-Refine, a two-stage framework that transfers offline SAM 3 pseudo-masks to a compact segmentation network. Six fixed prompts are fused by an unweighted 50% vote to stabilize the teacher output. The student first learns the foreground silhouette and is then refined with signed-distance-field, skeleton, and area objectives derived from the pseudo-mask. A sample-level gate, computed from prompt agreement, the valid-prompt ratio, and pseudo-mask area plausibility, reduces the influence of unreliable pseudo-geometry. On the SpaceSense-Bench HJM lockbox set, GeoDistill-Refine improves Image IoU and Boundary F1 by 0.0456 and 0.1380, respectively, over a plain pseudo-label student. External evaluations on the SPEED+ Lightbox and Sunlamp domains and on TANGO show competitive regional overlap together with gains in boundary quality or foreground precision. The deployed TinyUNet contains 0.263 M parameters and requires approximately 1.1 ms per image on an RTX 4090; SAM 3 pseudo-mask construction and the auxiliary geometry branches are used only during training.