Diamond tool manufacturing is a very traditional industry with long complex processes and several personalized customizations. There is great difficulty in quality control. In the implementation of corresponding machine vision, it is necessary to overcome the softness of customization, high precision of defect detection and complex image contents of quality inspection etc. This session will take the industry as an example to carry out product function design, technical implementation and business applications. A PC-based machine vision system is used for demonstration and explanation.Speaker(s):
The market for service robotics is experiencing a perfect storm of circumstances for growth, like market readiness, technology maturity, delivered value, cost, availability, and the maturity of related technologies. Generally, two or more SoC are used for service robotics; one for running robot operating system (ROS) to collect information on the environment and control motion, and the other(s) for hosting GUI to interact with customers. Due to the rich app ecosystem, Android is always selected as GUI OS, and Rosbridage or RosJava is used as the communication solution between ROS OS and Android OS.
However, this has led to customer complaints, such as high cost and maintenance, performance overhead due to TCP/IP based comminution and increased risk of DDoS attacking on some core nodes of ROS. This session will focus on our novel solution, RosRoid, which will reduce the pain point by consolidating Ros OS with Android in a single computing platform. The two mixed OS run in single kernel and isolated from each other. Moreover, binder based communication solution reduces performance overhead and brings security check mechanism to avoid DDoS attack.Speaker(s):
As robust depth cameras become more affordable, many new products will benefit from true 3D vision. This presentation will highlight the benefits of depth sensing for tasks, such as autonomous navigation, collision avoidance and object detection in robots and drones. We will explore a fully functional SLAM pipeline built using free and open source software components and an off-the-shelf Intel® RealSense™ camera, and observe how it performs in real-time environment mapping and tracking.Speaker(s):
GeoVision is a DSS player in the Top 30 global security companies. Come and see how we work with GeoVision using a set of Intel® Software Tools to optimize computer vision & Deep Learning workloads. Join us to learn the insights that can benefit your own development efforts.一起加入我們 來了解 GeoVision 全球30大的數位安全監控廠商是如何透Intel® Software Tools來最佳化他們的電腦視覺及深度學習軟體產品.Speaker(s):
关于数字标牌数字标牌（Digital Signage）是一种异军突起的新概念媒体，具体所指的在大型商超、便利店、地铁、酒店、机场等人流汇聚的公共场所，通过不同类型的终端显示设备发布商业、时政、财经、娱乐以及其他定制信息的多媒体视听系统。数字标牌强大的适应性使其可以不同的形态部署在特定的场所，并在指定的时间段对特定人群进行信息的指向性推送。随着物联网技术的发展，数字标牌正迅速展现出日益强大的传播能力，被誉为继平面媒体、电台、电视以及因特网之后的“第五媒体”。大数据驱动零售业智慧转型随着数字标牌技术的发展，行业竞争日趋激烈，各类智慧数字标牌设备层出不穷，但就目前来看零售业仍然处于向智慧化转型的进程中。人工智能产品设备及解决方案供应商-杰和科技认为，在已经到来的物联网时代，“数据”才是推动行业发展的引擎。万物互联，其本质就是数据的连接，通过智慧数字标牌终端所收集到的数据，经由网络在云端进行汇总和分析，从而转化为驱动行业发展的强大动力。所以，杰和科技认为，加速零售业的智慧转型，重点就在于高效地收集和利用智慧零售的精髓即“大数据”，针对这一趋势，杰和科技在其自主研发的“杰和数字标牌信息发布管理系统”（GDSM）中整合了适用于智慧零售行业的终端信息收集以及云端大数据汇总分析能力。零售大数据从何而来实体零售数据从何而来？杰和科GDSM通过整合计算视觉技术（Computer Vision），利用整合在数字标牌中的摄像头采集并识别顾客信息，这包括但不限于顾客属性数据（年龄层、性别）、顾客身份信息（会员）、客户光顾频次、顾客行走轨迹、门店热点区域以及顾客的停留时间。当杰和GDSM应用于智能货架时，摄像头不但可以采集顾客信息，还可以采集到货架上货物的实时存量、客户拿起/放下的产品信息以及客户在货架前滞留的时长。当GDSM应用于电子试装镜/电子试衣镜时，也能采集到零售商所需的数据，如顾客的年龄和外貌画像、最受欢迎以及最不受欢迎的产品款式、顾客的搭配喜好等。智慧零售时代，数字标牌的选址无论是从目标客户定位、市场环境定位还是店铺定位，都需要通过线下的数据分析，进行整体的、全新的智能规划。因此GDSM所适配的数字标牌播放器除了适配高性能摄像头以外还集成了GPS模块和温湿度感应器，结合地理位置反馈数字标牌周遭环境和顾客流量。数据的汇总与积累可以预见，通过整合在数字标牌中的摄像头和传感器，前端设备每天会收集海量的数据，这些数据或将暂时储存于前端设备中，并在网络传输宽裕的时间段上传到GDSM云端以便进行集中分析。大数据是靠时间积累起来的，如果积累的样本不足，对数据分析的准确性就会产生影响。因此GDSM将不断整合前端感应设备，持续丰富前端大数据的收集种类。只有将所有数据集中分类，并不断积累和筛选，才能为大数据的有效分析利用和人工智能的正确反馈提供坚实的数据支持。数据的实时查看与智能分析智慧零售时代，实时数据监测为过程管控提供了更有效的手段。简单来说，假如企业在做一系列的大型促销活动，想要第一时间获取到促销效果的反馈并借此对后续活动进行合理调整，那么实时数据就是一个很重要的支撑。杰和科技认为GDSM利用数字标牌收集前端数据的意义在于，可以通过数据的精准收集，实现大数据的积累，从而驱动实体零售业的智慧化转型和企业的长期发展带来准确可靠的大数据基础。Speaker(s):
Computer Vision (CV) solutions often experience bandwidth, storage and latency limitations. Scaling CV applications to meet Power-and-Performance (PnP) KPIs and cost requirements can be expensive, and deploying AI models on different hardware may require a complete solution redesign. By experience, it is a non-trivial process to efficiently integrate deep learning capabilities into traditional CV applications without significant performance overhead. To address these topics, Intel has developed optimized AI Frameworks, Intel® System Studio suite and Intel® Distribution of OpenVINO™ Toolkit mainly targeted for embedded visual computing developers.
In this presentation, we will be covering the following topics: Deep Learning training using Intel® Optimization for TensorFlow. Using the common inference API for heterogeneous execution using Intel’s Deep Learning Deployment Toolkit (DLDT) for Intel processors (for CPUs), Intel® Processor Graphics (for GPUs), and vision processing units (VPUs), as well as Intel optimized OpenCV and OpenVX libraries for traditional computer vision. Develop and profile CV applications natively on the target platform using Intel® System Studio, which is a cross-platform tool suite purpose-built to simplify system bring-up, and improve system and application performance on Intel platforms.Speaker(s):
The majority of African population are still using charcoal and firewood to cook due to the high cost of LPG. How about using Internet of Things (IoT) to tackle this issue? Yes, LPG adoption has traditionally been difficult for customers who live in regions that lack modern energy infrastructure, where cash flow is constrained. Because of this, most of these households make small daily purchases of wood or charcoal for cooking. However, through an IoT-enabled smart meter, families can use their smartphones to prepay for gas using mobile wallets - that is the power of IoT. This transaction activates the valve to supply the purchased amount of gas, allowing customers to pay as for the LPG as they cook with the flexibility of an incremental purchase. This compliments the earnings volatility common among low-income households, especially those who work in the informal sector. The constant monitoring of the data allows remote scheduling and delivery of a new LPG cylinder before the customer runs out of gas, assuring a consistent supply of quality LPG gas.Speaker(s):
Legacy assets worth millions of dollars, which cannot be discarded, have been one of the major factors hindering the adoption of IoT by several organizations. Most of these systems were not designed to "talk" to each other or the internet which makes it really difficult to obtain data from them. This session will focus on how computer vision can be leveraged, across industries, to obtain data and infuse IoT features to monitor and control legacy systems.Speaker(s):
Alleantia provides the most scalable, quick solution for implementing Industrial IoT, for any industrial sector and environment. Alleantia IIoT edge software easily and intuitively connects machines, production lines, industrial equipment, from any brand, through a library of 5000+ device drivers for plug and play machine integration. The software exposes open APIs supporting on-premise or on-cloud Industry 4.0 applications, including Machine Learning, Predictive Maintenance, AR/VR, etc. APIs available include MQTT, SQL flavours, Azure IoT Hub, AWS IoT, IBM Watson, REST and many others. Alleantia & Advantech implement a unique co-strategic partnership to integrate Alleantia Industrial IoT edge into Advantech reference IoT Gateways and industrial computing and with Advantech Wise-PAAS IoT application development platform. The Plug & Play IIoT RFP Ready Kit is one of the first products, using the powerful, rugged UTX3117 workhorse, supporting many connectivity requirements and perfectly operating Alleantia IIoT software on Linux or Windows IoT OS.Speaker(s):
In today’s physical retail, inventory distortions in the form of out-of-stocks, overstocks, and non-compliance with planograms (product placement plans) cost retailers and suppliers a staggering 1.75 trillion USD loss in annual revenue worldwide. As a countermeasure, stores constantly task their personnel to frequent manual checks at the shelves to replenish them when necessary. But the sheer diversity, amount and pace of fast-moving goods makes this manual process very overwhelming, untimely and error-prone, and arguably a waste of valuable human resources given the current state of AI and IoT technologies. Shelf integrity inspection is almost entirely visual, allowing for its automation using Computer Vision.
We propose the Vispera ShelfSight, a shelf-monitoring and management system that can analyze retail shelves for out-of-stocks, planogram compliance and many other KPIs in real-time. ShelfSight uses IoT cameras mounted on shelf aisles, and is powered by state-of-the-art deep learning algorithms developed by Vispera. Our deep models are customized for specific visual tasks such as detecting-recognizing product facings, detecting-reading price tags, tracking customers and more, and operate with almost perfect accuracy on edge/near-edge enabling real-time execution. ShelfSight can provide retailers instant access to the digitized view of their store and give them an operational excellence with to-the-point store actions.Speaker(s):