开箱即用的动漫头像检测包
基于Yolo V3的动漫头像检测模型,仅需200张标注数据即可在多种番剧、线描、背面等场景下实现较高检出率。开源代码和模型权重,支持开箱即用。
Category
Articles filed under Computing.
基于Yolo V3的动漫头像检测模型,仅需200张标注数据即可在多种番剧、线描、背面等场景下实现较高检出率。开源代码和模型权重,支持开箱即用。
DigitalOcean server incidents and migration to Azure—site layout may be temporarily affected during the transition.
Creating gigapixel panoramas with Leica SL and vintage R lenses: from Seattle skyline to macro photography of watches and dollar bills, with tips for successful stitching.
基于Caffe的斗图机器人:用深度学习提取表情特征,在数据库中找最近邻实现自动回复,效果意外地好但也暴露了ImageNet预训练模型在二次元表情上的局限。
微信机器人功能升级:群标签云统计、个人发言分析、活跃度可视化,并接入ElasticSearch和Kibana实现实时监控,算法从TF升级到TF-IDF。
探讨聊天机器人的核心价值:利用聊天记录做跨群转发、主题归并、脉络梳理和统计分析,让碎片化信息变得有组织,提供比单纯推送更独特的价值。
Build a Raspberry Pi-based air quality monitoring system with gas and dust sensors, plus a DIY air purifier using a box fan and MERV 13 filter.
Using a FLIR One thermal camera to identify home energy leaks: window drafts, heater inefficiencies, and door gaps. Practical DIY approach to thermal debugging.
The IFTTT delay API now supports passing extra POST parameters (Value1, Value2, Value3), enabling more complex automation recipes like delayed blog cross-posting and smart home scene scheduling.
智能家居进阶:加入智能温控、August门锁、iBeacon定位和Amazon Echo,探索从遥控到自动化再到智能化的发展路径,IFTTT作为统一控制中枢。