石鑫华视觉论坛

 找回密码
 注册会员
查看: 52|回复: 0

[MVTEC] MVTEC MERLIC 5.5.2 Linux x64-linux/aarch64-linux完整版和运行版的扩展MERLIC 5.5.2 Examples for the runtime version

[复制链接]
  • TA的每日心情
    郁闷
    3 小时前
  • 签到天数: 3389 天

    连续签到: 11 天

    [LV.Master]2000FPS

     楼主| 发表于 2024-10-31 16:19:41 | 显示全部楼层 |阅读模式 来自:广东省东莞市 电信

    注册登陆后可查看附件和大图,以及购买相关内容

    您需要 登录 才可以下载或查看,没有账号?注册会员

    x
    MVTEC MERLIC 5.5.2 Linux x64-linux/aarch64-linux完整版和运行版的扩展MERLIC 5.5.2 Examples for the runtime version

    文件名: merlic-5.5.2-examples.tar.gz
    文件大小: 678133387 字节 (646.72 MB)
    修改日期: 2024-07-08 20:43
    MD5: ba744d386fac0a44c8fc373674deeeb9
    SHA1: 916345cd20fc9ed809f8706373eddf5fedf30d57
    SHA256: 6b569ee0fb804b23dcda43d565e8ee3b3039d6de267aaf9b9041d84b011e5d28
    CRC32: ec3657f0

    百度网盘下载:
    通过百度网盘分享的文件:merlic-5.5.2-examples.tar.gz
    链接:https://pan.baidu.com/s/1jQxCzHmJojyI2j-1Yv9zZQ
    提取码:
    游客,如果您要查看本帖隐藏内容请回复

    --来自百度网盘超级会员V7的分享

    MERLIC 5.5 的最新功能
    在新版本中,我们延续了 MERLIC 的既定路线——简单的流程集成与强大的机器视觉方法相结合。这就是为什么 MERLIC 5.5 再次包含一个界面,进一步简化 MERLIC 的流程集成。还将有令人兴奋的新功能,

    最新改动?
    为了改进流程集成,开发了一种新的 TCP 套接字插件,可以与不支持复杂协议的设备进行基于文本的通信。 新技术包括用于对大量物体进行计数的基于深度学习的深度计数方法,以及用于识别颜色的新工具。 另一个新功能“高动态范围(HDR)成像”可以处理高对比度差异并有效消除图像中曝光过度或曝光不足的区域。

    新的 TCP 套接字插件


    MERLIC 5.5 通过引入新的 TCP 套接字插件扩展了其连接功能。 该插件能够使用高度可定制的基于文本的协议。 用户现在可以配置特定的、简单的 ASCII 消息,其中仅包含与其应用程序相关的信息。 这些消息可以由不支持 OPC UA 等复杂协议的设备轻松处理。 该插件可以在 MERLIC 的 RTE(运行时环境)设置中轻松配置,为将 MERLIC 集成到现有系统提供了简单的解决方案。

    深度学习计数 Deep Counting

    MERLIC 5.5 中新的“深度学习计数”概念工具允许用户高效、准确地计数许多物体,同时检测它们的位置。 与其他基于深度学习的方法不同,该功能可以用最少的标记快速训练。 这简化了流程并显着提高了对象计数的效率,特别是在涉及可变形材料或散装货物的情况下。 同时,这也减少了通常与深度学习模型训练相关的时间和成本。

    新的颜色识别工具

    MERLIC 5.5 引入了一种处理颜色的新方法。 经过训练后,新的“识别颜色”概念工具就可以在各种条件下可靠地检测颜色。 用户甚至可以通过设置可接受偏差的特定阈值来进一步提高检测精度。

    该概念工具非常适合多个应用组合,例如零件验证或选择、验证正确的电缆连接或确认正确电阻器的安装,有助于简化和增强质量控制流程。

    高动态范围 (HDR) 成像

    MERLIC 5.5 包含 HDR(高动态范围)成像的新方法。 此功能扩展了现有的“合并图像”工具,现在配备了 HDR 功能,可将不同曝光的图像合并为单个 HDR 图像。 此增强功能可确保处理高对比度差异,有效消除图像中曝光过度或曝光不足的区域。 因此,即使是最具挑战性的照明条件也可以准确地表示和分析。

    这为各种应用开辟了新的可能性,包括反射表面的测量和增加图像的动态范围。

    Release Notes for MERLIC 5.5.2
    This document provides the release notes for MVTec MERLIC 5.5.2 , as released in July 2024.
    Contents
    • Supported Operating Systems
    • MERLIC Creator
    • Tools
    • Licensing
    • Miscellaneous
    • Known Issues

    Supported Operating SystemsWindows
    MERLIC 5.5.2 is available for Windows 10 (64-bit) or newer.
    Please refer to the "Readme" in the MERLIC manual for more information about the system requirements.
    Linux
    MERLIC provides a test version for evaluating MERLIC RTE (Runtime Environment) for Linux on PC (Linux x86_64) and for Arm-based platforms (AArch64). This means that both embedded devices based on the widely used Arm architecture and Linux alternatives to the typical Windows operating system can now be tested for image processing with MERLIC. This is a first step towards making MERLIC available on additional platforms in the future.
    Developing the MVApp is still done on a Windows system with the MERLIC Creator. The application can then be tested on the "new" systems using MERLIC RTE for Arm-based platforms and Linux. Cameras can be set up remotely with the help of MERLIC RTE Setup. The test version requires a specific license. Please contact MVTec to get more information about how to obtain this license.
    MVTec appreciates any feedback on the test version. The feedback can be submitted on the MERLIC feedback website which can be accessed from the MERLIC Creator either via the feedback icon on the top right or via the "Help" menu.

    MERLIC Creator
    • Fixed Problems
      • MERLIC crashed when a tool with a graphic display was deleted and the deletion caused a race condition or block in the MERLIC Vision App. This problem has been fixed.

    Tools
    • Fixed Problems
      • When using the tool "Read Text and Numbers with Deep Learning" and moving the mouse pointer over the text in the image, the tooltip providing further information on the read text was not displayed anymore. This problem has been fixed.
      • When using the concept tool "Segment Image Pixel Precise" and moving the mouse pointer over the image to see the detected class labels, the respective tooltip providing further information on the assigned class and confidences was not displayed anymore. This problem has been fixed.

    Licensing
    • Enhancements
      • Some virus protection suites falsely warned that a component of MERLIC may be malicious. This problem has been fixed. The license protection tools have been updated.

    Miscellaneous
    • In some cases, loading MVApps with a large number of nested "Branch on Condition" tools or complex tool flows took very long because of the long duration required for checking the Tool Flow graph. This problem has been fixed.

    Known Issues
    The following issue is already known and in development process. It will be fixed for upcoming releases.
    • In frequent cases, when using TensorRT™ accelerated processing units in MERLIC tools with deep learning technology, MERLIC crashes if the required memory for the deep learning model exceeds the available memory on the GPU.
    • When an instance of MERLIC Creator or MERLIC RTE of version 5.0 is remotely configured using a MERLIC RTE Setup 5.1 or newer, removing an image source from a configuration will cause the Creator or RTE application to crash. We recommend to upgrade Creator/RTE to the current version or, in lieu of that, to use RTE Setup 5.0 to configure Creator/RTE 5.0.
    • When an instance of MERLIC Creator or MERLIC RTE of version 5.2 (or newer) is remotely configured using a MERLIC RTE Setup 5.1, it is not possible to rename image sources or configurations, despite the fact that this feature has been added in MERLIC 5.1. We recommend upgrading the MERLIC RTE Setup to version 5.2.
    Legal Notes :
    Modicon® is a registered trademark of Schneider Electric USA, Inc.
    Intel, the Intel logo, OpenVINO, the OpenVINO logo, and Movidius are trademarks of Intel Corporation or its subsidiaries.
    回复

    使用道具 举报

    您需要登录后才可以回帖 登录 | 注册会员

    本版积分规则

    LabVIEW HALCON图像处理入门教程(24.09)
    石鑫华机器视觉与LabVIEW Vision图像处理PDF+视频教程11种全套
    《LabVIEW Vision函数实例详解2020-2024》教程-NI Vision所有函数使用方法介绍,基于NI VISION2020,兼容VDM21/22/23/24

    QQ|石鑫华视觉论坛 |网站地图

    GMT+8, 2024-11-27 11:24

    Powered by Discuz! X3.4

    © 2001-2024 Discuz! Team.

    快速回复 返回顶部 返回列表