蓝带 拱桥 区别
This series will cover Cordatus AI platform in detail. Please feel free to ask any question in the comments.
本系列将详细介绍Cordatus AI平台。 请随时在评论中提出任何问题。
目录 (Table of Contents)
什么是Cordatus AI？ (What is Cordatus AI?)
Cordatus is an AI platform for desktop PCs, workstations, servers, NVIDIA Jetson and other CPU-based edge devices. It enables you to easily create and start using your favorite Machine Learning/Deep Learning/Computer Vision environments such as Tensorflow, PyTorch, OpenCV, DLib and many more without any installation. The platform offers NVIDIA CUDA-X software stack for NVIDIA GPU enabled systems and specific software optimizations for CPU-based servers and edge solutions by utilizing Docker containers. This enables both AI developers and beginners to start experimenting easily with ready to run and state of the art Computer Vision applications created by FAIR, NVIDIA and AI community and provided by Cordatus AI platform.
Cordatus是一个AI平台，适用于台式机，工作站，服务器，NVIDIA Jetson和其他基于CPU的边缘设备。 它使您无需安装即可轻松创建和开始使用自己喜欢的机器学习/深度学习/计算机视觉环境，例如Tensorflow，PyTorch，OpenCV，DLib等。 该平台为使用NVIDIA GPU的系统提供了NVIDIA CUDA-X软件堆栈，并通过利用Docker容器为基于CPU的服务器和边缘解决方案提供了特定的软件优化。 这使AI开发人员和初学者都可以轻松地开始试验，这些试验由FAIR ， NVIDIA和AI社区创建，并由Cordatus AI平台提供，可以立即运行和使用最新的计算机视觉应用程序。
支持的平台和操作系统信息 (Supported Platforms and OS Information)
For PC/Workstation/Servers:
对于PC /工作站/服务器：
- Ubuntu 18.04 Ubuntu 18.04
For NVIDIA Jetson Nano/TX2/Xavier NX/AGX Xavier:
对于NVIDIA Jetson Nano / TX2 / Xavier NX / AGX Xavier：
- JetPack 4.3 JetPack的4.3
- JetPack 4.4 JetPack 4.4
辅助功能 (Accessibility)
Cordatus is a highly accessible platform where you can manage your devices and containers from both web browser and dedicated application. In order Cordatus to be able to communicate with your devices, Cordatus application needs to be installed and running on each of your devices. The application itself is available as Debian package (.deb) for download and it can be installed by following the official Cordatus Documentation.
Cordatus是一个易于访问的平台，您可以在其中通过Web浏览器和专用应用程序管理设备和容器。 为了使Cordatus能够与您的设备通信，需要在每个设备上安装并运行Cordatus应用程序。 该应用程序本身可以Debian软件包(.deb)的形式下载，可以按照官方Cordatus文档进行安装。
The dedicated application can be used as both GUI and CLI application. CLI scenario is applicable when it comes there is no display output with remote servers or edge devices such as NVIDIA Jetson platforms.
专用应用程序既可以用作GUI应用程序，也可以用作CLI应用程序。 CLI场景适用于远程服务器或边缘设备(例如NVIDIA Jetson平台)没有显示输出的情况。
Cordatus网站 (Cordatus Web)
Cordatus Web enables users to manage all of their connected devices from a single web page. It also can be used to create, delete, start and stop all environment or application containers remotely. The web page also features monitoring where you could get detailed real-time usage metrics such as CPU Usage, CPU Heat, GPU VRAM Usage, GPU Utilization, GPU Heat and RAM Usage from your connected devices.
Cordatus Web使用户可以从一个网页管理所有连接的设备。 它还可以用于远程创建，删除，启动和停止所有环境或应用程序容器。 该网页还具有监控功能，您可以在其中获取所连接设备的详细实时使用指标，例如CPU使用率，CPU热量，GPU VRAM使用率，GPU利用率，GPU热量和RAM使用率 。
For the NVIDIA Jetson platform, you can also switch between Jetson’s performance modes within this web page!
对于NVIDIA Jetson平台，您还可以在此网页中在Jetson的性能模式之间切换！
Cordatus Web also supports AI environment/application container management for remote devices where you can easily deploy, remove, start, stop your AI environments/applications and create public URLs for your Jupyter Notebooks runs within your containers.
Cordatus Web还支持针对远程设备的AI环境/应用程序容器管理，您可以在其中轻松部署，删除，启动，停止AI环境/应用程序，并为在容器中运行的Jupyter Notebooks创建公共URL。
Cordatus GUI / CLI应用程序 (Cordatus GUI/CLI Application)
Cordatus AI platform offers a dedicated application to establish communication between your devices and Cordatus itself. There are two approaches for the dedicated application. First one, where you have a direct display output from an adapter like HDMI, DisplayPort, etc. you can start the Cordatus GUI Application and;
Cordatus AI平台提供了专用的应用程序，可在您的设备与Cordatus本身之间建立通信。 专用应用程序有两种方法。 第一个是从HDMI，DisplayPort等适配器直接获得显示输出的位置，您可以启动Cordatus GUI应用程序；
- Manage (create, delete, start, stop) all your AI environment/application containers 管理(创建，删除，启动，停止)您所有的AI环境/应用程序容器
- Create public Jupyter Notebook URLs for your AI environments/applications 为您的AI环境/应用程序创建公共Jupyter Notebook URL
Start a fast and TensorRT-optimized object detection with Cordatus Inference Engine.
使用Cordatus推理引擎启动快速且TensorRT优化的对象检测。
- Get raw stream from physically connected cameras (USB, CSI-MIPI, GMSL2, FPDLINKIII) 从物理连接的摄像机(USB，CSI-MIPI，GMSL2，FPDLINKIII)获取原始流
- Get real-time usage metrics of the device currently in use, get system information (OS information for PCs and JetPack information for NVIDIA Jetsons) 获取当前正在使用的设备的实时使用情况指标，获取系统信息(PC的OS信息和NVIDIA Jetsons的JetPack信息)
- Switch between performance modes for NVIDIA Jetsons 在NVIDIA Jetsons的性能模式之间切换
- List all Docker images on your system 列出系统上的所有Docker映像
The second use case with the Cordatus dedicated application is where you don’t have a direct display output from the device such as remote servers on AWS, Azure, GCP, etc. or NVIDIA Jetson deployed on a drone or an autonomous machine. In this scenario, the application can be started from a terminal without any GUI output and enables the device to connect to Cordatus Web.
Cordatus专用应用程序的第二个用例是，您没有来自设备的直接显示输出，例如AWS，Azure，GCP等上的远程服务器或部署在无人机或自动机上的NVIDIA Jetson。 在这种情况下，可以从没有任何GUI输出的终端启动应用程序，并使设备能够连接到Cordatus Web 。
引入Cordatus AI环境 (Introducing Cordatus AI Environments)
Cordatus offers optimized Machine Learning/Deep Learning/Computer Vision development environments within Docker containers. This eliminates the need to install ML/DL/CV libraries from scratch and possible optimization problems thus lets developers to adapt their main focus on the application development itself.
Cordatus在Docker容器内提供了优化的机器学习/深度学习/计算机视觉开发环境。 这消除了从头开始安装ML / DL / CV库的需要，避免了可能的优化问题，从而使开发人员可以将主要精力放在应用程序开发本身上。
All of the environment containers includes all the necessary software stack and optimizations. For instance, “Tensorflow by Cordatus” environment offers both CPU and GPU optimized versions where CPU optimized one includes CPU-specific optimizations like parallelization libraries, chip specific build procedures, etc. and the GPU optimized one includes NVIDIA CUDA and cuDNN packages (TensorRT included within Application containers).
所有环境容器都包括所有必需的软件堆栈和优化。 例如，“ Tordorflow by Cordatus”环境提供了CPU和GPU优化版本，其中CPU优化版本包括并行化库，特定芯片构建过程等CPU特定的优化，而GPU优化的其中包含NVIDIA CUDA和cuDNN软件包(包括TensorRT)在应用程序容器中)。
带Cordatus的Nvidia GPU云容器 (Nvidia GPU Cloud Containers with Cordatus)
Cordatus offers all the popular Machine Learning/Deep Learning library containers from Nvidia GPU Cloud (NGC) which are optimized by NVIDIA. This allows users to deploy and run NGC containers on their NVIDIA GPU enabled PCs, workstations and servers using a web browser, Cordatus GUI App or Cordatus CLI App (Note that GUI App is not available for remote servers).
Cordatus提供Nvidia GPU Cloud(NGC)提供的所有流行的机器学习/深度学习库容器，这些容器已由NVIDIA优化。 这使用户可以使用Web浏览器，Cordatus GUI应用程序或Cordatus CLI应用程序(请注意，GUI应用程序不适用于远程服务器)在支持NVIDIA GPU的PC，工作站和服务器上部署和运行NGC容器。
Note: A GPU with at least Pascal architecture required in order to use the NGC containers.
注意：要使用NGC容器，至少需要具有Pascal架构的GPU。
引入Cordatus AI应用程序 (Introducing Cordatus AI Applications)
Cordatus AI platform provides optimized ready-to-run state of the art AI applications created by Facebook AI Research (FAIR), NVIDIA and open-source AI community. These enables AI developers to customize state of the art AI applications and proven models to fit for their own needs and also enables beginners to get the best hands-on experiences and insights related to AI application development path.
Cordatus AI平台提供了由Facebook AI Research(FAIR)，NVIDIA和开源AI社区创建的优化的，可立即运行的先进AI应用程序。 这些使AI开发人员能够自定义最先进的AI应用程序和经过验证的模型，以满足他们自己的需求，还使初学者可以获得与AI应用程序开发路径相关的最佳动手经验和见解。
There are various AI applications within Cordatus that are ready to run without installing or building anything. For example; Detectron2 from FAIR packs object detection, human pose estimation, semantic segmentation, instance segmentation and panoptic segmentation in one application.
Cordatus中有各种AI应用程序，它们可以在不安装或构建任何东西的情况下运行。 例如; 来自FAIR的Detectron2在一个应用程序中打包了对象检测，人体姿势估计，语义分割，实例分割和全景分割。
引入Cordatus推理引擎 (Introducing Cordatus Inference Engine)
Cordatus AI platform has also it’s own inference engine called Cordatus Inference Engine for both x86 (PCs, workstations, Server, etc.) and arm64v8 (NVIDIA Jetson) based systems. CIE ships with CUDA-X software stack (CUDA, cuDNN, TensorRT) as well as TensorFlow and ONNX and leverages all of these libraries to provide fast and accurate object detection results. Cordatus Inference Engine currently supports:
Cordatus AI平台还拥有自己的推理引擎，称为Cordatus推理引擎，可用于基于x86(PC，工作站，服务器等)和arm64v8(NVIDIA Jetson)的系统。 CIE随附CUDA-X软件堆栈(CUDA，cuDNN，TensorRT)以及TensorFlow和ONNX，并利用所有这些库提供快速，准确的对象检测结果。 Cordatus推理引擎当前支持：
- Object Detection 物体检测
- Face Detection 人脸检测
- Covid-19 Face Mask Detection Covid-19面罩检测
This optimized inference engine utilizes the following models;
该优化的推理引擎利用以下模型；
- YOLO V3 (yolov3-tiny-288, yolov3-tiny-416, yolov3-288, yolov3-416 and yolov3-608) YOLO V3(yolov3-tiny-288，yolov3-tiny-416，yolov3-288，yolov3-416和yolov3-608)
for object detection (/w TensorRT),
用于物体检测(/ w TensorRT)，
- MT-CNN 神经网络
for face detection (/w TensorRT) and
用于面部检测(/ w TensorRT)和
- Face Mask Detection Model 面罩检测模型
for Covid-19.
适用于Covid-19。
All of these applications only takes a few clicks from cordatus.ai or Cordatus GUI App to start and deploy them on your both local and remote PCs, workstations, servers and NVIDIA Jetsons.
所有这些应用程序都只需从cordatus.ai或Cordatus GUI App上单击几下，即可在本地和远程PC，工作站，服务器和NVIDIA Jetsons上启动和部署它们。
翻译自: https://medium.com/cordatus-ai-platform/cordatus-ai-17a1c3b0e26a
蓝带 拱桥 区别
4271
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
