Unverified50% confidenceFactExact time
MAE (Masked Autoencoder) was proposed by He et al. in 2021; it randomly masks a high proportion of image patches (such as 75%), feeds only the visible patches into the encoder, and then a lightweight decoder reconstructs the masked pixels.
1
Sources
50%
Confidence
Long-term
Relevance
11/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedMAE（Masked Autoencoders）通过遮盖图像块并重建来学习视觉特征71% similarUnverified掩码自编码器（MAE）通过随机遮盖图像块后让模型重建，可用于雷达图像的自监督预训练68% similarUnverifiedStable Diffusion combined with ControlNet can be used to generate hidden artistic QR codes that look like illustrations but remain scannable.64% similarUnverifiedMasked editing is a local image modification technique in which the user provides a binary mask and the model regenerates only the area covered by the mask while preserving the rest at the pixel level.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1121923API
curl https://kongchang.com/api/v1/knowledge/claims/1121923MCP
get_claim(id=1121923)