Unverified50% confidenceFactExact time
Depthwise separable convolution was popularized by Google in MobileNet and decomposes a standard convolution into two steps: a depthwise spatial convolution and a 1×1 pointwise channel-mixing operation.
1
Sources
50%
Confidence
Long-term
Relevance
9/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedS2地理分区系统由Google开发，基于希尔伯特空间填充曲线将球面递归细分为层次化的四叉树结构67% similarUnverifiedGoogle's PoseNet is a lightweight model designed specifically for the browser side, based on the MobileNet architecture.64% similarUnverifiedGoogle DeepMind explored the possibility of controlling mobile devices through the Android Accessibility API.62% similarUnverified谷歌采用端云协同的混合架构：简单任务在本地完成，复杂任务无缝上传至云端处理62% similarUnverifiedGoogle open-sourced a lossless compression algorithm designed for inference that can cut inter-cluster transmission bandwidth by half.61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/837583API
curl https://kongchang.com/api/v1/knowledge/claims/837583MCP
get_claim(id=837583)