Embedded vision developers embrace neural networks

February 03, 2020 //By Peter Clarke
Embedded vision developers embrace neural networks
A survey conducted by the Edge AI and Vision Alliance shows a sharp jump in the use of neural networks for computer vision applications in 2019.

The percentage of active users of neural networks went up to 81 percent in 2019 from 59 percent in 2018. The percentage using or planning to use neural networks hit 94 percent.

The most popular platforms to run those neural networks on remain CPUs and GPUs although their use is diminishing in favour of dedicated computer vision processors and dedicated deep learning processors, the survey reveals.

The survey revealed that TensorFlow remains the most popular software framework for designing, training and evaluating neural networks for vision tasks. However, it's popularity is declining rapidly in favour of almost all other approaches including Caffe, Caffe2, Matlab and most notably in 2019 Mxnet.

The survey was conducted among 705 computer vision developers across a range of industries, organizations, geographical locations and job types in October 2019. The Edge AI and Vision Alliance used to be known as the Embedded Vision Alliance.

Related links and articles:

www.edge-ai-vision.com

Download computer vision developers survey

News articles:

Groq rocks neural networks

Neural style image transfer

GrAI Matter, Paris research gives rise to AI processor for the edge

AI processor startup emerges from Imagination

Greenwaves preps GAP9 IoT Processor on FDSOI


Vous êtes certain ?

Si vous désactivez les cookies, vous ne pouvez plus naviguer sur le site.

Vous allez être rediriger vers Google.