Machine vision which software is better (machine vision software ranking)

1, Halcon, VisionPro as the representative of the traditional machine vision software, by calling a variety of algorithms factors, for the target characteristics, customized settings of various parameters. It is good at solving the quantitative problems mainly based on measurement, and the simple problem of determining the presence or absence, but for the needs of complex detection class, the leakage rate / false detection rate is high. Although the price of the software has been reduced again and again, the domestic software selling price of only a few thousand dollars or even free, but the pre-development and later maintenance costs are high, in the use of facing great limitations.

2, Cognex's VIDI, Sualab (which has been acquired by Cognex) as the representative of the deep learning software, the deep learning of the annotation, training, testing process, packaged in the way of PC software, reducing the threshold of the user to use deep learning technology. It mainly solves the complex defect detection problems that cannot be solved by traditional vision, or improves the development efficiency by using the annotation method of deep learning. However, from the practical application effect, the current deep learning is mainly based on supervised learning, for different scenarios, requires a large number of model selection, tuning parameter and other work, can not be solved for all scenarios, generic, therefore, does not solve the traditional algorithms need to be customized to solve the problem of the development of traditional algorithms and the user's use of the threshold is high, the ordinary engineers of the tuning parameter of the depth of the deep learning is generally no basis.

3, Beijing Moment Vision Intelligent Technology Co., Ltd, founded in 2017, has received nearly 10 million yuan in financing, innovatively provides a cloud-based industrial vision deep learning tools, online to achieve the labeling, training, testing process, relying on a large number of different scenes accumulated in the cloud deep learning model library, users upload, labeling pictures, will be automatically matched by the cloud with the optimal deep learning model After users upload and label pictures, the cloud will automatically match the optimal deep learning model and parameters for training, and users do not need any operation, and after testing the effect on the line, they can download the SDK to run locally, which is really a general-level deep learning tool for industrial vision. At the same time in the commercial, Moment View Intelligence will cloud tools directly for free, download the sdk, only need to buy Lisence dongle, can be run locally, is committed to deep learning technology applied to all industrial vision scenes.