WiMi Hologram Cloud Inc (WIMI)

[StockpickerCAPC]

WiMi Is Training An AIHolo Robot To Generate 3D Holograms.

Source
https://www.newstrail.com/wimi-is-training-an-aiholo-robot-to-generate-3d-holograms/

December 14, 2022

WiMi Hologram Cloud, Inc. (NASDAQ:WIMI) is training an AIHolo robot to generate 3D holograms for users using AI artificial intelligence in neural networks, which can go through the instructions in the training prompts and provide detailed responses. It can automatically generate multiple holograms for the user’s requirements for the hologram, giving the user various holograms that are understood according to the customer’s requirements. WiMi AIHolo robot is based on the open source OpenAI-Photo open-source image generation basis and intelligent recognition of local items, depth of field, depth, phase, etc. of the image, and then controlled by the complementary difference algorithm for 3D holographic image pixel high-precision processing.

The initial stage uses the user’s feedback to strengthen the training AIHolo. The engineer will use human supervision to train an initial AIHolo model, give instructions and provide practice subjects. AI training engineer is the user of the AI hologram generation program AIHolo. In the back end, the AI training engineer can access the model and write the program to help AIHolo generate the images accordingly.

Then the engineer will build optimization models, create optimization logic for reinforcement learning and collect comparison data. At least two or more models ranked by quality are needed, so there will be multiple AI training engineers to give instructions, collect different comparison data and different sample results and then compare them and rank them manually so that models are optimized. And then, engineers will consider using the PPO (Proximal Policy Optimization) to fine-tune the model and iterate the optimization model many times.

AIHolo may also have many problems and limitations because AI is a computer program. New areas are required to complete the initial training model and design the framework for future training techniques. AIHolo also needs to avoid responding to sensitive and harmful instructions or generating biased 3D holograms, which requires review to stop such unsafe content. Of course, there may be some false positives and missing critical information, which requires the following collection of user feedback to help AIHolo improve and optimize the system’s work. It is also very likely that the user needs to provide more complex instructions so that AIHolo makes further inquiries or guesses through its understanding of the user’s intentions and cannot get the best results. Therefore, subsequent user participation is needed to train and optimize AIHolo and continuously iterate.

The launch of AIHolo will likely disrupt the entire 3D holographic mapping field. It may change some of the ways 3D holographic mapping engineers used to work and improve the efficiency of enterprises. WiMi AIHolo will give users and enterprises the convenience and variety of options to obtain 3D holograms using AI artificial intelligence of neural networks.