SeamCam: Quantifying Seamless Camouflage via Multi-Cue Visual Detectability

2026-05-15

AI Summary

This study addresses the lack of standardized quantitative evaluation for seamless animal camouflage in existing methods. The authors formulate camouflage assessment as a category-conditioned visual localization problem and propose SeamCam, the first computable metric that quantifies an animal’s visual detectability within a known category by jointly optimizing object detection and instance segmentation masks—higher scores indicate stronger camouflage. They also introduce CamFG-1.5k, a high-quality benchmark dataset free from occlusion bias, and pioneer the use of SeamCam as a preference signal in Direct Preference Optimization (DPO) for diffusion models. Experiments demonstrate that SeamCam achieves 78.82% agreement with human judgments, outperforming current approaches by approximately 25%, and significantly enhances the realism of camouflage images generated by diffusion models.

Abstract

Animals are described as effectively camouflaged when they blend seamlessly with their surrounding, yet no standardized quantitative measure of this seamlessness exists. We address this gap by framing camouflage evaluation as a visual localization problem: a well-camouflaged animal is one that remains difficult to detect even when its category is known. We introduce SeamCam (Seamless Camouflage), a metric that quantifies how detectable an animal is from the available visual evidence. Given an image and a target species, SeamCam generates category-conditioned detection proposals, extracts segmentation masks, and identifies the subset whose collective union yields the highest IoU with the ground-truth mask. The SeamCam score is one minus this maximum recoverable localization signal, where a higher score indicates stronger camouflage (i.e., lower detectability). In a human two-alternative forced-choice study with 94 participants and 2,390 comparisons, SeamCam achieves 78.82% agreement with human camouflage difficulty judgments, outperforming state-of-the-art by about 25%. We then demonstrate SeamCam's utility as a preference signal for Direct Preference Optimization (DPO) to fine-tune a diffusion-based inpainting model for camouflage generation. This offers an affordable training approach with an objective explicitly suited for camouflage generation, unlike typical diffusion models. To support rigorous benchmarking, we further introduce CamFG-1.5k, a curated dataset of 1,521 high-resolution images in which animals are fully visible prior to camouflage generation, enabling unbiased evaluation by controlling for occlusion artifacts present in existing datasets. https://7amin.github.io/SeamCam/

Problem

• seamless camouflage
• visual detectability
• camouflage quantification
• animal camouflage
• visual localization

Innovation

• SeamCam
• visual detectability
• camouflage quantification
• Direct Preference Optimization
• CamFG-1.5k