NI Vision Development Module 2019 x86/x64 Final
Dành cho người đã có một số hiểu biết khái niệm cơ bản trong Vision như ROI (region of interest), cách lập trình LabVIEW, lấy hình ảnh từ camera với Vision Development Module. Hướng dẫn này chỉ nhằm mục đích mở tài liệu đúng chỗ & định hướng cho người quan tâm cụ thể ứng dụng OCR, còn cụ thể step-by-step thì đã có rất chi tiết trong tài liệu Help rồi.
Vision Development Module (VDM) is designed to help you develop machine vision applications using LabVIEW or C/C++ and deploy those applications to Windows or NI Linux Real-Time hardware. With its comprehensive function library, you can access hundreds of image processing algorithms and machine vision functions to enhance images, check for presence, locate features, identify objects, measure parts, and more. VDM also includes the Vision Assistant, which is an algorithm engineering tool that simplifies vision system design by helping you develop algorithms for deployment on either CPUs or FPGAs. Use the Vision Development Module to take an open and adaptable approach to machine vision software development. You can choose the right hardware for your application and configure cameras, acquire images, and analyze inspection results to build fully customized machine vision systems.
New Features in Vision Development Module 2019:
-Deep Learning OpenVINO™ Support
-Added new Deep Learning API (Under Machine Learning > Deep Learning) to support loading of OpenVINO™ models, Intermediate Representation (IR) format. The new VI added is IMAQ DL Model Create (OpenVINO).
-Added support for executing the OpenVINO™ IR models.
-Added Model Optimizer Support using python. Python scripts for converting models from Deep Learning frameworks like Tensorflow and Caffe to IR are provided. (More details about Model Optimizer and it's usage are available here: %NI_MO_INSTALL_PATH%\Readme.md and %NI_MO_INSTALL_PATH%\Readme.html).
-The OpenVINO™ Inference APIs are supported on Linux RT, Windows.
-Added support for Xilinx® Kintex® UltraScale™ FPGA IPs capable of processing 8 pixels in parallel, using the following VIs:
-IMAQ FPGA AutoBThreshold ICV ROI U8x8_DRAM256x2
-IMAQ FPGA AutoBThreshold ICV U8x8_DRAM256x2
-Video Support Improvements
-AVI Updates: Support for writing AVI files larger than 2 GB.
-New Codecs Installed: VP8, VP9.
-Support for parallel Match Pattern instances with the same template image.
-Template Editor improvements
-The Template Editor window is now resizable.
-Template masking improvements - Masks can be modified and not committed immediately, new magic wand tool to mask areas.
-New Auto Mask feature, detecting the object and masking the background automatically.
OS:Windows 7 SP1 / Windows 8.1 / Windows 10 (x86-x64) Windows Embedded Standard 7 (SP1) / Windows Server 2008 R2 (SP1) 64-bit / Windows Server 2012 R2 64-bit
CPU:Pentium 4/M or equivalent
Memory:512 MB RAM
Display Assistant:1,152 × 864 resolution video adapter with a 24- or 32-bit display
Display for others:1,024 × 768 resolution video adapter with a 24- or 32-bit display
NI Vision Development Module 2019 Bug Fixes:
-Extract HOG outputs different Feature vector at every call.
-Vision examples are not linked correctly through CVI Example Finder.
-Deep Learning Examples were not searchable through LabVIEW Example Finder.
-In some cases, IMAQ Match Pattern 4 can crash due to a race condition.
-Find Edge VI does not have an output that returns whether an edge was found.
-Calling a VI that has Vision functions in a TestStand sequence with Adapter Run-Time engine mode with Auto detect using VI version selected will sometimes cause the UI to close and LabVIEW to silently crash if there are multiple LabVIEW Run-Time engines installed..
-Writing to AVI files is limited to a 2GB size.
Pass : benhvientinhoc.com