Unverified50% confidenceFactExact time
FlowNet于2015年首次用卷积神经网络端到端估计光流，RAFT于2020年通过全对相关体积并迭代优化大幅提升光流估计精度
1
Sources
50%
Confidence
Long-term
Relevance
8/24/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedFlux由Black Forest Labs于2024年推出，采用流匹配（Flow Matching）训练范式替代传统DDPM扩散过程62% similarUnverifiedFlow Matching was proposed by Lipman et al. in 2022, extending discrete time steps to continuous-time manifolds and replacing stochastic sampling with a deterministic ODE solver.61% similarUnverifiedSince 2020, diffusion models have comprehensively surpassed GANs (Generative Adversarial Networks) in image generation quality.55% similarUnverified2022年前后，以ChatGPT、Stable Diffusion为代表的产品引发了当前的生成式AI浪潮55% similarUnverifiedFlow is an AI creation platform that integrates open-source models such as Stable Diffusion and Flux as well as various commercial models, and offers a one-month free trial period.53% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/794860API
curl https://kongchang.com/api/v1/knowledge/claims/794860MCP
get_claim(id=794860)