Unverified50% confidenceFactExact time
SegFormer is a lightweight Transformer segmentation framework proposed by NVIDIA Research in 2021, whose core innovation is replacing the traditional ViT with a hierarchical Mix Transformer (MiT) encoder and adopting an All-MLP decoder.
1
Sources
50%
Confidence
Long-term
Relevance
10/8/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedToday's mainstream visual encoders are highly similar at the underlying architecture level, with almost all being based on Transformers.72% similarUnverifiedChangeFormer采用层次化Transformer编码器，在LEVIR-CD和DSIFN-CD等基准上取得了SOTA性能68% similarUnverifiedLaya pairs a bidirectional ModernBERT-large encoder with a Transformer head trained from scratch that scores [MASK] option tokens.67% similarUnverifiedTransformer架构最初由Vaswani等人于2017年为NLP设计，后被迁移到视觉（ViT）、音频、蛋白质结构预测等几乎所有AI子领域67% similarUnverifiedThe Transformer architecture proposed in 2017 laid the foundation for modern machine translation.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/987193API
curl https://kongchang.com/api/v1/knowledge/claims/987193MCP
get_claim(id=987193)