NEWS · MODELS · #726
LoRA adaptation of DINOv3 ViT greatly improves SAS ATR; later refinements give no further gain
The paper adapts a DINOv3 Vision Transformer to underwater synthetic aperture sonar (SAS) automatic target recognition using a three-stage, parameter-efficient pipeline: Stage 1 applies Low-Rank Adaptation (LoRA) with a frozen backbone, Stage 2 adds hard-negative mining, and Stage 3 uses Supervised Contrastive Learning (SupCon). On at-sea SAS data with a mission-level geographic split and comparisons at 85% test recall repeated over three seeds, LoRA (rank 4) raised AUPRC from 0.300 to 0.679 ± 0.027 while training only 0.26% of weights; the hard-negative and SupCon stages showed no meaningful improvement versus matched controls.
KEY POINTS
- The paper adapts a DINOv3 Vision Transformer to underwater synthetic aperture sonar (SAS) automatic target recognition using a three-stage, parameter-efficient pipeline: Stage 1 applies Low-Rank Adaptation (LoRA) with a frozen backbone, Stage 2 adds hard-negative mining, and Stage 3 uses Supervised Contrastive Learning (SupCon).
- On at-sea SAS data with a mission-level geographic split and comparisons at 85% test recall repeated over three seeds, LoRA (rank 4) raised AUPRC from 0.300 to 0.679 ± 0.027 while training only 0.26% of weights; the hard-negative and SupCon stages showed no meaningful improvement versus matched controls.
- Shows that a single parameter-efficient LoRA adaptation can bridge natural-image pretrained ViTs to scarce underwater acoustic data and produce large gains, while stacked supervised refinements may be unnecessary.
WHY IT MATTERS
Shows that a single parameter-efficient LoRA adaptation can bridge natural-image pretrained ViTs to scarce underwater acoustic data and produce large gains, while stacked supervised refinements may be unnecessary.