Visual Management Based on Integrated Convolution Neural Network

Using convolutional neural network technology for model compression, the convolutional layer operation speed is improved through multiple row-column parallel computing units in the convolution operation acceleration module. At the same time, it is proposed that a convolution operation acceleration module should be used to build a multi-way parallel architecture of the convolution layer, saving time and improving management efficiency. By utilizing IP core development and efficient management of algorithm hardware, image big data pre-processing and post-processing calculations are realized through PS-side instruction scheduling. Using a dedicated neural network processor architecture and a dedicated algorithm acceleration core, computing power support is provided, the autonomy of the visual image management system is improved, and the dependence on measurement and control equipment is reduced, thereby realizing the rapid deployment of neural networks.