Unverified50% confidenceFactExact time
The VTF900 dataset includes expression, pose, and occlusion variations, among which pose and occlusion are the most common sources of failure for cross-spectral recognition in real-world scenarios.
1
Sources
50%
Confidence
Long-term
Relevance
10/20/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedCoverage analysis compares the target virtual camera trajectory with recorded viewpoints to determine which regions are supported by real pixels and which require model extrapolation, and is the core mechanism for assessing hallucination risk.67% similarUnverifiedPerceptual Loss uses a pre-trained VGG network to extract high-level feature representations of images, computing the gap between generated images and target images in feature space rather than pixel space.66% similarUnverifiedVLM 视觉方案（如 WebVoyager、SeeAct）存在幻觉问题、坐标定位精度不足，且每次操作需消耗大量 token 处理图像信息64% similarUnverifiedMulti-view stereo (MVS) accuracy depends on texture richness and view coverage, and tends to fail in weakly textured, reflective or occluded regions.63% similarUnverifiedQuantization introduces precision loss, which is particularly pronounced in precision-sensitive areas such as high-frequency textures, facial details, and lighting transitions.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1074842API
curl https://kongchang.com/api/v1/knowledge/claims/1074842MCP
get_claim(id=1074842)