Feed Entities Podcasts Deep Dives Developers About

[KongchangAI]

Real-time AI knowledge base. Every claim is traceable to its source, with data freshness under 24 hours.

Powered by 240+ first-hand AI sources

Knowledge

Feed
Hot Entities
Podcasts

Developers

CLI
MCP Server
REST API
About

© 2026 KongchangAI. The freshest, most accurate AI knowledge

Unverified50% confidenceFactExact time

语义分割方法用于车道检测存在计算成本过高的问题

1

Sources

50%

Confidence

Medium-term (~90 days)

Relevance

7/2/2026

First Seen

Valid until: 9/30/2026

Sources

Deep Learning Lane Detection: From Segmentation to Classification for 300+ FPS Real-Time Inference

bilibili论文发刊罗小黑6/12/2026

Related Claims

Unverified路由分类器需同时满足轻量（推理耗时毫秒级）和准确两个要求，误判会导致质量下降或资源浪费70% similar Unverified是否将未铺装的越野小径（track）纳入可达范围会显著影响难抵极点的计算结果67% similar Unverified路由决策若用小模型可能判断不准，若用大模型则路由开销会抵消节省，需要在准确性、延迟、成本之间权衡66% similar Unverified路由层一旦出错会导致调用错误专家，使下游每个答案都偏差，因此必须保持高精度62% similar Unverified路由层不压缩的工程依据是：路由错误会在整个前向传播中级联放大且无法后续纠正，而路由层参数量占比极低62% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/42105

API

curl https://kongchang.com/api/v1/knowledge/claims/42105

MCP

get_claim(id=42105)