Intelligent clipping solution

The Qiniuyun intelligent editing solution supports the structural analysis of massive audio and video materials, applies the algorithm capabilities such as "face detection and recognition, voiceprint comparison, facial expression recognition, and lens type" to achieve the function of intelligent editing through the consistency comparison of audio and pictures, and supports the import of mainstream editing software (such as Premiere) for secondary fine cutting.

Scheme function

Voiceprint recognition

The speaker's voice features are extracted for identity verification. The voice information of each channel of audio is extracted through the voiceprint model, and then search and compare in the voiceprint database to locate the identity of the guest in the audio file.

Face recognition/detection

Based on the advanced facial analysis technology, it has realized the functions of face detection, feature extraction and search, and accurately recognized the face information of the guests in the corresponding picture in the video material.

Landscape differentiation

The type of shot is determined based on the statistical information such as the proportion of face and picture size in the picture, the number of heads, etc.

GPU virtualization

Self developed distributed AI services and GPU virtualization technology are applied to realize multi instance parallel computing, on-demand scheduling, support horizontal expansion, and maximize software and hardware performance.

XML Export

Based on the time dimension, output structured information such as voice print, face, lens, etc. for material analysis, realize the audio and picture synchronization editing function, and generate PR compatible XML files.

Mirror cutting burr removal

For scenes with frequent shot transitions, the clip switching logic is targeted to optimize, and the "mirror cutting burr removal algorithm" is applied to make the shot transition more smooth and further improve the visual experience.

Scheme architecture

Application scenarios

reality show

The reality show has added a large number of elements such as editing and commentary, and its super concurrent processing ability can process more than 100 routes of stand materials at the same time, greatly improving the efficiency of staff.

variety show

With the interaction between guests and audience, the editing complexity of multiple audio tracks and video tracks has increased, and the time from shooting to completion can be greatly shortened through fast pre editing.

Post production of film and television

It provides software and hardware technical services for massive audio and video raw materials, seamlessly connects with editing software, significantly shortens editing time, and quickly completes various complex production tasks.

Scheme advantages

Low threshold

Support public cloud/private cloud, pure software/hardware integration and other product delivery forms, and significantly reduce the threshold for video editing.

automation

From material preparation to automatic plywood, and then to automatic editing, there is no need for manual intervention in the whole process to help pre editing business.

high efficiency

From the end of recording to the beginning of creation, the cycle is greatly shortened from 4-5 days to less than 6 hours, saving time and cost.

Leading algorithm

Self developed distributed AI service and GPU virtualization technology are applied to realize multi instance parallel computing and on-demand scheduling. It has industry-leading AI capabilities such as face recognition, voiceprint contrast and scene recognition.

WeChat consultation

Telephone consultation

Intelligent customer service