中文/EN
WeChat Bilibili Zhihu Weibo

AI Video Enhancement

Based on deep learning technology and training with massive video data, AI video enhancement is a way to realize video image quality restoration, color enhancement and clarity improvement by leveraging resolution enhancement, denoising, HDR and frame rate up-conversion technologies, so as to improve the visual sensory quality of videos for users. AI video enhancement products are applicable to broadcast media, online video and other application scenarios, and can be used for intelligent image quality improvement and secondary application of some video contents, such as old movies and TV dramas. Combining with NEOLINK's IDC infrastructure construction and operation capacity, we provide customers with "AI+ cloud computing" solutions for flexible deployment.
Products

Core Functions

  • Super Resolution
    Artificial intelligence algorithms are adopted to reconstruct low resolution videos into ultra HD resolution videos with 4K standard format index (3840x2160), so as to enhance the details of images and enrich the image texture.
  • HDR
    By upgrading the standard dynamic range (SDR) to high dynamic range (HDR), expanding the narrow gamut (BT.709) to wide gamut (BT.2020), and converting 8-bit images to 10-bit images, it can achieve all-round image quality improvement, and vividly display the colorful details in the image.
  • Frame Interpolation
    In view of the attention mechanism, we calculate the bidirectional optical flow motion information between video frames, adopt intelligent frame interpolation strategy for different video contents, and increase the frame rate from 25p to 50p, thereby significantly improving the smoothness of videos, and creating ultimate playback experience.
  • Denoising
    By constructing the channel-space hybrid attention mechanism to distinguish the noises with different distortion degree, and then learning the distortion characteristics efficiently, the core AI algorithms can solve different types of distortion and restore clearer texture details.

Core Advantages

  • All-round Intelligent Algorithm Enhancement
    According to different video enhancement requirements, the intelligent super-resolution enhancement algorithm can improve video quality from five dimensions (including spatial resolution, dynamic range, color gamut, frame rate and gray resolution), and achieve multidimensional balance, thereby delivering ultra HD videos for better visual experience.
  • Flexible Deployment
    A variety of deployment methods are supported, including private deployment, cloud deployment and hybrid cloud deployment, etc. Customers can select the optimal deployment programs according to different scenarios. Customers can deploy a set of private video enhancement services by using the integrated software and hardware system or video enhancement services through cloud API. Distributed computing technology is adopted to provide customers with high-performance, cost-effective cloud-based AI video enhancement services.
  • High Performance
    Distributed technology is adopted for computing cluster management. The computing tasks can be decomposed into multiple nodes for parallel processing, thereby improving the performance of video processing. Meanwhile, nodes can be added dynamically at any time, thereby increasing the computing power of the whole cluster.
  • High Availability
    With distributed deployment of application environment, dynamical expansion can be achieved through load balancing, and the monitoring alarm module can monitor the health status in real-time manner, thereby ensuring highly available system services.

Application

Repair of Old Videos


The restoration of an old movie consists of a series of processes, such as color matching, image restoration, and restoration quality appraisal. At present, manual restoration is widely used. Generally, it takes several weeks, several months, or even as long as half a year to restore a movie. The restoration costs range from several hundred thousand to several million yuan, and it is not easier than making a new movie.


When using the AI video enhancement service, you can import the original video file into the system, and set parameters of the output video (resolution, HDR, frame rate, etc.). Subsequently, the system will automatically complete the original video detection, frame splitting, enhancement, and synthesis processing, and make use of NEOLINK's AI computing network, and perform enhancement calculation through distributed clusters, so as to efficiently complete the video image enhancement.



Online Videos


With the rapid development of online video industry, video content has become the key to success. In addition to copyright procurement, video quality enhancement has also become an important ability for video production platforms.


AI video enhancement service, which can be accessed through API, can significantly improve the AI video enhancement capability of video production platforms. In addition to quick access, it minimizes the damages to the original production process, and reduces the impact on the entire business process.