Concentration, Not Uncertainty: Why Targeted Synthetic Data Doesn't Help Camouflaged Object Detection
Camouflaged object detection requires pixel-accurate masks, but obtaining such annotations is slow and costly, making synthetic training images an attractive alternative.
Key points
- Under a fixed generation budget, however, it remains unclear which real-image regions to target for synthetic data generation.
- We study an uncertainty-guided generation strategy that clusters the unlabelled real images, identifies clusters on which the model is least certain, allocates synthetic generation toward those clusters, and iteratively retrains the model.
- Across 103 training runs, uncertainty-based targeting does not outperform random allocation.
- Separately, we find substantial data contamination in CHAMELEON, with 50 of its 76 images duplicated from training data despite the standard overlap check reporting zero overlap.
Sources (1)
- [1]Concentration, Not Uncertainty: Why Targeted Synthetic Data Doesn't Help Camouflaged Object DetectionarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 7, 10:27 AM
Camouflaged object detection requires pixel-accurate masks, but obtaining such annotations is slow and costly, making synthetic training images an attractive alternative.
Under a fixed generation budget, however, it remains unclear which real-image regions to target for synthetic data generation.
Extractive summary: sentences quoted from the sources.