Search papers, labs, and topics across Lattice.
This paper introduces a model-agnostic saliency-depth conditioning strategy to enhance zero-shot segmentation of communication-tower components in UAV imagery, addressing the challenge posed by cluttered backgrounds. By combining appearance-based saliency with monocular relative depth, the authors construct a coarse prior that effectively suppresses irrelevant structures, leading to improved localization. The proposed methods, SD-Grounded-SAM and SD-SAM 3, significantly outperform existing models on the TOW-300 dataset, with SD-SAM 3 achieving the highest instance-segmentation performance and SD-Grounded-SAM minimizing false positives.
Zero-shot segmentation can be dramatically improved in cluttered environments by integrating saliency and depth information, leading to fewer false positives and better localization of communication towers.
Fine-grained segmentation of communication-tower components in UAV imagery is essential for automated inspection, yet task-specific models are hard to develop due to limited instance-level annotations. Zero-shot segmentation models offer a promising alternative, but in cluttered scenes, visually similar background structures interfere with component localization, causing missed instances and false positives. We propose a model-agnostic saliency-depth foreground-conditioning strategy combining appearance-based saliency with monocular relative depth to construct a coarse tower prior and suppress irrelevant content. We integrate this module with Grounded-SAM and SAM 3, yielding SD-Grounded-SAM and SD-SAM 3. SD-Grounded-SAM further applies geometric and depth-aware box refinement before mask generation, while SD-SAM 3 relies on SAM 3's internal setup. On TOW-300, a dataset of 340 communication-tower UAV images, our strategy improves both baselines: SD-SAM 3 achieves the strongest instance-segmentation performance, while SD-Grounded-SAM produces fewer false positives. Ablations confirm complementary gains from saliency, depth, and box refinement, improving robustness in cluttered scenes.
