Core Technology

Imperial Vision is deeply involved in the field of AI vision with leading analysis ability and understanding of images and videos in this industry.
We pursue the ultimate in image quality enhancement & restoration technology and video image analysis technology and provide full-stack intelligent vision solutions as well as services to customers.

AI Restoration Model

Utilize over 1000 restoration models to provide ndustry-leading AI enhancement & restoration technology of image quality.

AI Restoration Algorithm

Incorporate more than 20 AI restoration algorithms, being at the leading edge in many fields such as reconstruction effect and reconstruction speed in the international arena.

AI intellectual property

Possess more than 150 patents related to artificial intelligence
Video Super Resolution
Features: Boost video resolution up to 8K, make video to be vividly, realistic and natural.While enhancing visual effects, ensuring the coherent of generated textures and less jittery.
Application: Movies, TV series, documentaries, animation, TV shows and short videos on mobile phones, etc.
Detail Enhancement in Video
Feature: The algorithm of detail enhancement uses the AI model to extract features from the image, and adaptively adjusts the detail information to achieve the effects of removing defocusing blur, enhancing details and sharpening edges to further improve the subjective image quality.
Application: Mainly used for post-processing of denoising and super-resolution, if the input has a lot of noise or compressed pseudo-shadow, it will be slightly amplified.
Face Enhancement in Video
Features: Face enhancement in Video combines a variety of AI algorithms, which can automatically detect the position, appearance and size of faces in the video, and restore and enhance the face to improve the clarity.
Application: Applicable for characters in videos with half-length portraits. Can be applied to most video types, se.g. short videos taken by mobile phones, online videos or video conferences, etc. Currently limited by the size and angle of the human face, can olny improve faces facing the camera will be optimized, and the number of faces can not exceed five. Interlaced video is currently not supported.
Upconversion of SDR Video
Features: Utilize AI algorithm to convert SDR video to HDR10 or HLG video, improve the dynamic range, color gamut, bit depth of video, output better image effect.
Application: All SDR videos.
Down Conversion of HDR
Features: Utilize AI algorithm converting HDR video to SDR video, ensure minimum distortion of image.
Application: All HDR10 or HLG videos.
Video Frame Interpolation
Features: Using AI algorithm to generate intermediate frames at any time between two frames, increasing the video frame rate to a maximum of 120fps, improving smoothness, reducing stuttering and improving visual perception.
Application: Reproduction of high frame rate video or production of slow-motion video. Not suitable for videos with extremly violent motion.
Video Denoising
Features: Based on a multi-frame video denoising algorithm, our intelligent denoising can remove noise and grain while preserving as much realistic detail and texture in video as possible and reduce flicker and other pseudo-shadow which could be caused by frame-by-frame noise reductionsolution.
Application: Movies, TV series, documentaries, animation, TV shows and short videos taken by mobile phones, etc.
Coloring Black-and-White Videos
Features: AI can automatically analyze different areas in videos to select the best color for them, calculate the differences among pictures in different time segments, and render the coloring results to each frame of the video naturally.
Application: For videos with few scene switching and stable motion. Not suitable for videos with extremely violent motion.
Destratch of Videos
Features: The automatic scratch removal algorithm can automatically identify the location of scratches and restore them. The scratch removal algorithm can remove the flickering scratches from pictures and make them look like cleaner.
Application: Videos with flickering scratches.
Watermark Removal of Video
Features: The watermark removal algorithm is a deep learning method that uses temporal and spacial information of videos to complete the watermark obscured content. The results obtained from the deep learning based watermark removal method are more natural and smooth.
Application: For all types of video with logo, watermark, ads, etc.
Removal of Flicker
Features: The flicker-removal algorithm can eliminate the flicker phenomenon by smoothing the brightness of multiple frames, so that solve the flicker problem of the video.
Application: Videos with flicker problem.
Features: The de-jitter algorithm is a traditional image method that obtains the jitter-parameters of the picture according to the similarity of the images between frames and adjusts the images according to the parameters, which can produce a better video effect.
Applications: For videos with shaky pictures due to shooting problem
Style Conversion of Video
Features: Convert objects and background in video into japanese manga style or american comics or other art style.
Application: All kinds of videos with theme about landscape, people and animals.
Separation of Chracters
Features: Apart characters or specified objects from background automatically by using algorithm.
Application: Changing the background in videos, photos, pictures, etc.

Setting new highs in the industry for many times

Imperial Vision Technology is the first company in the world to utilize AI for video restoration and enhancement. Our super-resolution technology has achieved the best results in international public data for many times. We have been recognized by over 100 corporate client including Intel, ARM, CCTV, China Telecom , Fuyao Group etc.

  • 2017July
  • 2017December
  • 2018December
  • 2019August
  • 2020October
  • 2021May
  • 2022February
  • 2017July

    Our pioneering SRDenseNet super-resolution technology was presented at the 2017 IEEE International Conference on Computer Vision (ICCV) and is currently peer cited over 900 times.

  • 2017December

    Nvidia Incepetion World 2017 "China AI Startup Challenge" Top 5 Comapny

  • 2018December

    Won the first prize of 5th China Youth Innovation and Startup Competition in startup category. Invited to attend Asia Conference on Computer Vision and presente our newly developed super-resolution technology "Image Super-Resolution Using Knowlege Distillation".

  • 2019August

    Won the Award of Excellence of innovateFPGA Design Contest Grand Final Award. More than 400 outstanding innovative teams from 11 countries around the world participated in the contest. We successfully advanced to the top 12 teams in the finals.

  • 2020October

    Won the the RTE2020 Algorithm - Image Super-resolution Challenge.

  • 2021May

    Won the first prize of the MediaAIAC (Radio, Television and Network Audiovisual AI Application Innovation Competition) for ultra-high definition video intelligent restoration

  • 2022February

    Won the first prize of Fujian Provincial Science and Technology Progress Award. The project of "R&D and Industrialization of Key Technologies for Ultra-HD Video Intelligent Processing" jointly completed by Division Technology, Fuzhou University and Fujian Radio and Television Group was awarded the Fujian Provincial Science and Technology Progress Award. first prize

Algorithm customization

As long as you need it, we will customize an exclusive AI image algorithm library for you.

Consult now

Pre-sales advice 400-006-5266
Business Cooperation
Talent recruitment