Unverified50% confidenceFactExact time
经典CNN依赖局部感受野逐层提取从纹理到形状的层次化特征
1
Sources
50%
Confidence
Long-term
Relevance
7/9/2026
First Seen
Sources
Block-Sparse Featurizers: Opening the Black Box of Vision Model Neural Geometry
hackernewshackernews7/7/2026
Related Claims
UnverifiedConvolutional Neural Networks (CNNs), inspired by the structure of the visual cortex, extract local features by sliding convolutional kernels across images, with the parameter sharing mechanism significantly reducing computational load.74% similarUnverifiedCNN通过局部感受野和权重共享机制高效提取图像的空间层级特征，是YOLO等检测模型的早期基础73% similarVerifiedIn 2019, researchers at the University of Tübingen discovered that CNN models pre-trained on ImageNet rely more on texture features than shape features when recognizing objects, a phenomenon known as texture bias.71% similarUnverifiedNeural style transfer measures style similarity by computing the Gram matrix of feature maps at each layer, enabling the separation of content and style.71% similarUnverifiedThe concept of style transfer can be traced back to a paper published by Gatys et al. in 2015, which used feature representations from convolutional neural networks (CNNs) to separate the content and style of images68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/313859API
curl https://kongchang.com/api/v1/knowledge/claims/313859MCP
get_claim(id=313859)