《人工智能视觉在制造业应用ppt课件.pptx》由会员分享，可在线阅读，更多相关《人工智能视觉在制造业应用ppt课件.pptx（21页珍藏版）》请在三一办公上搜索。
1、,AI History,人工智能的过去,人工智能发展的两个主要阶段Two main stages of AI,物理符号系统PSSH(Physical Symbol System Hypothesis),左侧：赫伯特西蒙 右侧：艾伦纽厄尔Left : Herbert A.Simon Right : Alan Newell,2012年以前：专家系统Before 2012 : Experts Definition,3,2012年以后：深度学习After 2012 : Deep Learning,4,人工智能发展的两个主要阶段Two main stages of AI,人工智能技术 AI Techno
2、logies,自然语言理解Natural Language Processing,计算机视觉Computer Vision,5,人工智能三要素 Three Elements of AI,反馈Feedback,数据3.Data,模型2.Model,人工Artificial,智能Intelligence,深度学习计算框架DL computing framework,计算力1.Computing power,人工智能大脑AI Brain,6,活的人工智能让产品更强大Live AI make products better,7,深度学习训练Deep Leaning Training,现象 Phenom
3、enon 1,000,001,结果 Result,?,8,The challenges of AI,人工智能的挑战,Artificial Intelligence,AI,大量的人工高昂的成本A lot of manual effortsHigh labor costs,10,准确率的提升Improvement Of Accuracy,准确率Accuracy,成本Cost,100%的准确率难度极大The accuracy of 100% Impossibly Difficult,11,Industrial AI,智能制造,传统工业视觉 vs 人工智能Traditional Industrial
4、Vision vs AI,传统工业视觉Traditional Industrial Vision,简单外观检测 Simple appearance detection,测距 Distance measurement,13,传统工业视觉缺陷Traditional Industrial VisionLimitations,固定位置（Fixed position）固定环境（Fixed setting）固定算法（Fixed algorithm）,14,拍摄条件要求低（Low shooting requirements）异常扩展性好（Good expansibility in exception）,人工
5、智能优势Artificial IntelligenceAdvantages,传统工业视觉 vs 人工智能Traditional Industrial Vision vs AI,人工智能能做什么？ What can AI do?,分类-什么Classification : What ?,Plant Identification x AI Deep Learning,形色,10000 Species x 98% x 30 million,15,植物识别 x 人工智能,检测-哪里Detection : Where ?,自动批改数学作业 x 人工智能Correct math homework x AI
6、 Deep Learning,98% x 9 million x 8000 Years,爱作业,16,人工智能能做什么？ What can AI do?,质检=什么+哪里Quality Assurance = What + Where,分类Classification,检测Detection,+,17,人工智能能做什么？ What AI can do?,智能制造应用案例Cases in Industrial AI,零件视觉索引管理Visual Index Management for Parts,配件管理（Parts Management）安全生产（Safety Control）,18,NO,Yes,配置检测Configuration Inspection,19,智能制造应用案例Cases in Industrial AI,复杂场景类裸眼质量检测Automation QA in Complex Scenes,20,智能制造应用案例Cases in Industrial AI,曲面检测Curved Surface Measurement,面临的挑战 Challenges :,划痕（Nicks）灰尘（Dust）多角度（Multi-angle）,21,智能制造应用案例Cases in Industrial AI,
