ROADMAP 2020/21 Next-gen goals

New tools for deep learning, 3D vision, embedded vision and much more will be on offer in the next year from machine vision suppliers

computer vision applications. Computer vision in low-cost

Euresys is designing the future of Coaxpress. For two years, Euresys and its subsidiary Sensor to Image have been working on Coaxpress-over- fibre, an interface for computer vision applications. The Coaxlink QSFP+,

which will launch at the beginning of 2021, will be the first acquisition card compatible with Coaxpress- over-fibre. Sensor to Image will provide Coaxpress-over- fibre bridge IP cores to enable camera and frame grabber manufacturers to implement the new standard. Notably, Euresys is co-operating with the Japan Industrial Imaging Association to ensure that this new concept becomes a commercially successful international standard adopted by the machine vision community. Euresys will also continue to

expand the capabilities of its Open eVision image analysis libraries with the release of deep learning libraries and 3D inspection functions. Finally, Open eVision will gain compatibility with Linux and additional embedded platforms.

In 2020, Irida Labs introduced EV Lib, a complete set of embedded vision functions based on deep learning and AI. In 2021, the company will offer EV Platform, a decentralised vision-as-a-service platform, providing end-to-end management of edge

and low-power edge devices has enabled the development of vision-AI applications running close to the data source. Following the hype of deep learning, there is now a strong demand for real-world edge vision applications. However, significant challenges need to be addressed to ensure scalability throughout the product life cycle, such as data management (resource- intensive data collection, sampling, labelling), problem formulation, and model design and update. EV Platform aims to

defragment the product development lifecycle, minimise the sensor acquisition period, annotation efforts and operation costs, automate the model update process, and provide a smooth experience across all deployed sensors. With less requirements on data campaigns, a much faster time-to-market can be achieved. EV Platform will also provide a unified management environment, analytics dashboards, and raw metadata, while supporting connectivity with a wide range of popular sensors, platforms and services (MCUs, Edge TPU, GPUs).

Michał Czardybon, CEO of Adaptive Vision, said that 2020 has been a year of major product releases for the company. Three new products have been introduced so far: firstly, Adaptive Vision Studio 5.0 with HMI events, parallel programming features, and a new Smart edition for end users; secondly, the Weaver inference engine for deep learning; and thirdly,, an online platform for image annotation and dataset management. Adaptive Vision will continue to focus on these three


products over the rest of the year. The development plan for Studio 5.1 is already fixed, Czardybon said, with the key new features being an offline mode strictly connected with image set management, and improved user experience for deep learning tools. After

Pleora will expand the plug-in machine learning capabilities of its AI Gateway product to help lower inspection costs. Alongside already released hyperspectral and inspection plug-ins, the embedded platform will be integrating additional deep learning-based classification, segmentation, and object detection capabilities. With the AI Gateway no-

code plug-ins, end-users and integrators can deploy

spending 2019 on bringing deep learning to a high level of performance, through the use of Weaver, the current focus is a return to accuracy and development of additional tools, he added. The company is starting

development of a public API for Zillin, which will allow customers to connect model training functionality to the online system for auto- labelling purposes.

Statement from Dr Olaf Munkelt, managing director of MVTec Software: ‘Looking to 2021, MVTec will advance its machine vision software and provide customers with new technologies, as well as improvements to core technologies. Halcon users can look forward to a new approach to OCR, and a new code reader, while Merlic customers will benefit from improved camera configuration and handling, enhanced PLC connectivity, and new image processing tools. ‘Making deep learning

more accessible for users and increasing its flexibility is also an important goal. On the software side, the MVTec deep learning tool

machine learning capabilities without any additional programming. Images and data are uploaded to user- friendly training software, with AI models then transferred and deployed on the AI Gateway in production environments. The platform’s Nvidia GPU

will continue to play an important role, receiving more functionality to integrate all steps of the deep learning workflow. In terms of hardware, MVTec is working on solutions which will allow customers to use dedicated AI accelerators for Halcon deep learning applications.’

@imveurope |

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