Roadmap 2019/20 2020 vision
Technology for deep learning, 3D imaging and embedded computing, along with new products for infrared imaging and illumination, are all on the horizon for the coming year
deep learning machine vision soſtware library – includes an improved one-class learning feature that detects defects without being trained on images of defective items. In addition, the label noise detection function is able to detect mislabelled images automatically, to give a more obvious dataset in training and get better accuracy. Looking toward the first half of next year,
Deep learning will continue to be a major trend for 2020, with soſtware and camera providers putting emphasis on neural network-based offerings. One of Adaptive Vision’s key focuses going into next year will be the development of a low-level deep learning framework, codenamed CoS. Only 20 to 50 samples will be needed to train the soſtware, which will be designed to give high performance on GPUs and CPUs. Deep learning soſtware tools are able to solve computer vision tasks through training, rather than programming. Another important point on Adaptive
Vision’s roadmap is working on its graphical development environment. Using the firm’s Inspector module, for example – a new edition of soſtware designed for simplified inspections – users do not have to create the entire structure of an application, but start with a predefined template, then focus on creating the vision pipeline. Te very first version of this, the minimal program view, is already available in the professional edition of the soſtware, 4.12.
www.adaptive-vision.com
In the coming months, Sualab plans to release a new version of SuaKit for smart camera applications. Te soſtware is designed so that deep learning algorithms can be embedded on smart cameras to avoid the use of a PC. Te current version, 2.3, of SuaKit – Sualab’s
SuaKit 3.0 will be released, providing improved usability and enhanced performance, say Sualab. Going forward, the firm will be looking for new vision projects that can make use of the features of SuaKit 3.0 and the smart camera version of the soſtware.
www.sualab.com
Euresys will expand the capabilities of its Open eVision image analysis libraries with the release of new deep learning and 3D functions. In addition to the classification function
already available in the Easy Deep Learning library, the firm is adding unsupervised
A statement from Dr Olaf Munkelt, managing director of MVTec Software: ‘In the first six months of 2020, MVTec will continue addressing the most important topics of machine vision with new releases of the Halcon and Merlic soſtware products, as well as its deep learning tool. Tese topics primarily include technologies such as 3D vision, deep learning, and embedded vision, with each becoming more important in the context of the Industrial Internet of Tings, or Industry 4.0. ‘MVTec’s soſtware releases will enable
users to apply AI-based technologies, such as deep learning, on a whole new level. MVTec will optimise its standard soſtware products with regard to embedded vision applications.’
www.mvtec.com
38 Imaging and Machine Vision Europe • Yearbook 2019/2020
defect segmentation and supervised semantic segmentation, both of which will be available by the end of the year. New object detection functionality, featuring labelled bounding boxes, will also be released early next year. A MIPI CSI-2 receiver IP core from Euresys’
subsidiary, Sensor To Image, will be available soon. It will be delivered with a reference design for fast development, and is compatible with Xilinx Artix7, Kintex7, Zynq7 and Ultrascale+ FPGAs.
@imveurope
www.imveurope.com
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52
