Efficient AI Platform for Image Processing
For the analysis of video streams using artificial intelligence (AI), normally extremely high performing hardware is required, which consumes a lot of energy. However, mother nature has proven to be an excellent teacher in this respect as well. Findings gained from brain research, which have led to the deciphering of processes associated with spatial vision, object recognition and movement identification in the cerebral cortex, are pointing the way to a method for highly efficient image processing. The “SlowFast” model is the result of emulating these processes on the basis of neural networks.
The implementation of this AI model has been analyzed at EDAG within the scope of research work whereby valuable findings have been gained for practical application. These help to better understand the influence exerted by various factors on the process of image recognition and thus enable targeted optimization of the relevant neural networks. The results are valid with respect to both the platform used, and as a means of improving different architectures of Convolutional Neural Networks (CNNs). This way, numerous image processing applications in the industrial field can be optimized, for example, in the area of quality control where scratches or air pockets need to be detected under a coat of paint.
Download our white paper now, titled “Efficient Image Processing using AI”, where the research work is described in detail.