{"slug": "impact-of-color-in-computer-and-human-based-object-recognition-under-occlusion", "title": "Impact of color in computer and human-based object recognition under occlusion", "summary": "A study posted to arXiv (2611.08808v1) found that human participants and most transformer models suffer reduced object recognition when occluders are colored to camouflage with the target object or scene, with scene-based camouflage hurting performance more than object-based camouflage. Classical convolutional neural networks showed the opposite trend, performing worst on neutral-colored occlusions, which the authors say suggests contextually related colors, especially scene-based ones, aid CNN recognition rather than disrupt shape or boundary cues. The work introduces three metrics — non-occlusion precision (NOP), non-occlusion recall (NOR) and activation in occlusion (AIO) — and reports that training augmentation with generated camouflaging occluders improved robustness to unseen camouflage for a selected CNN model.", "body_md": "arXiv:2611.08808v1 Announce Type: new \nAbstract: Partial occlusions challenge object recognition in both human and computer vision (CV), but the impact is largely dependent on the occlusion and model design. Properties relating to color are particularly important in biological recognition and frequently leveraged for camouflage, however there are still open questions about what makes them effective and how they translate to CV. This work studies partial occlusions that create camouflage, including those that blend in with target objects or with scenes, which alter object representations in distinct ways. Experiments evaluate recognition performance on both human participants and models ranging from convolutional neural networks (CNNs) to transformers, and additional analysis is performed on models to uncover trends in processing and object representations under multiple occlusion types. Novel metrics offer explainability of model behavior under occlusion: non-occlusion precision (NOP) and non-occlusion recall (NOR) measure alignment of model activations with and without occlusion, and activation in occlusion (AIO) measures the percentage of activations that occur in occluded regions. Findings show that human recognition, and most transformers, are negatively impacted by generated camouflaging occluders, with scene-based camouflage performing worse than object-based. Classical CNNs show a strong opposite trend, performing the worst on neutral-colored occlusions, suggesting that contextually-related colors, particularly scene-based, benefit object recognition rather than disrupting shape or boundary cues. Additionally, generated camouflaging occluders for training augmentation improves robustness to unseen camouflage for selected CNN model, and occlusion metrics suggest trends about learned mechanisms. Findings have interesting implications for alignment in human and computer vision, model vulnerabilities, and explainability.", "url": "https://wpnews.pro/news/impact-of-color-in-computer-and-human-based-object-recognition-under-occlusion", "canonical_source": "https://arxiv.org/abs/2611.08808", "published_at": "2026-11-09 05:00:00+00:00", "updated_at": "2026-11-09 05:24:05.149026+00:00", "lang": "en", "topics": ["computer-vision", "machine-learning", "ai-research", "neural-networks"], "entities": ["arXiv", "convolutional neural networks", "transformers"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/impact-of-color-in-computer-and-human-based-object-recognition-under-occlusion", "markdown": "https://wpnews.pro/news/impact-of-color-in-computer-and-human-based-object-recognition-under-occlusion.md", "text": "https://wpnews.pro/news/impact-of-color-in-computer-and-human-based-object-recognition-under-occlusion.txt", "jsonld": "https://wpnews.pro/news/impact-of-color-in-computer-and-human-based-object-recognition-under-occlusion.jsonld"}}