WiMi Hologram Cloud Inc (WIMI)

[StockpickerCAPC]

WiMi Hologram Cloud Develops Digital Image Processing Software To Improve Efficiency.

Source
https://www.newstrail.com/wimi-hologram-cloud-develops-digital-image-processing-software-to-improve-efficiency/

December 12, 2022

WiMi Hologram Cloud, Inc. (NASDAQ:WIMI), with holographic technology as its core, is dedicated to the algorithm research of image processing and imaging technology and has successfully developed holographic digital image processing software system based on visual saliency and channel attention mechanism to improve the image processing performance of its holographic intelligence system, which has been successfully applied in several fields or industries. Such as face recognition, AR/VR, 3D reconstruction, intelligent medical and smart city, medical devices, virtual reality products and services, industrial inspection, smart agriculture, intelligent robots, machine vision and intelligent security equipment, and high-end vision application systems, as self-driving cars.

The attention Mechanism is a data processing method in machine learning, which is widely used in different types of machine learning tasks such as natural language processing, image recognition, and speech recognition.
Channel Attention Mechanism and Visual Saliency can effectively improve the efficiency of image processing.
- (1) Channel-Attention Mechanism, using available features, select the most appropriate channel to extract the information of interest during image processing;
- (2) Visual Saliency, which analyzes the known features and simulates them through intelligent algorithms human visual characteristics to extract salient regions (i.e., critical regions of human interest) in images.
Attention mechanisms are beneficial in many vision tasks, such as image classification, target detection, semantic segmentation, video understanding, face recognition, person re-identification, action recognition, a small amount of display learning, medical image processing, image generation, pose estimation, super-resolution, 3D vision, multi-modal tasks, and self-supervised learning. The attention mechanism is similar to how people observe things outside. Generally speaking, when people watch external things, they first pay more attention to certain crucial local information that they tend to observe more and then combine the information from different regions to form an overall impression of the observed thing. The attention mechanism shifts the computer’s attention to the most crucial part.

The channel attention mechanism usually uses the SE Block model, essentially a channel-based Attention model that models the importance of each feature channel and then enhances or suppresses different channels for different tasks. The Channel Attention mechanism in computer vision is to learn different weights for each channel, in the channel dimension, with the same weight in the plane dimension. So channel domain-based attention is usually a direct global average pooling of information within a channel while ignoring the local information within each channel.

After the convolution operation, the pooling operation with global average pooling (GAP) is performed by the squeeze module, which compresses the spatial dimension with features, i.e., each two-dimensional feature map becomes an actual number with the same number of feature channels; then by the Excitation module, a two-layer hourglass-type structure is used (first descending and then ascending), which is achieved by a fully connected layer + Sigmoid function to accomplish the generation of weights for each feature channel, channel weights are learned to show the correlation between modeled feature channels, and finally, the obtained weight results are multiplied with the original feature map to present the display results. Applying the operational mechanism of this model to several benchmark models, a more significant performance improvement is obtained with a slight increase in computational effort. As a general design idea, it can be used for any existing network with substantial practical implications.

The channel attention mechanism is used to improve system performance by appropriately weighting features according to the importance of the input and simulating human vision for practical analysis and understanding of complex scenarios. The attention mechanism operates in a variety of ways, and WiMi’s R&D team is also conducting in-depth research in this area to improve the ability of holographic image processing systems to capture global information about holograms, refine image processing accuracy, improve computational efficiency, and reduce performance consumption.