WebKontrol Video Fingerprint is a compact representation of a video that contains a digital description of selected frames. It’s a small file created from a video, where each fingerprint is unique to the original video. When matched with other fingerprints, it enables rapid identification of copies of the full video or its parts.

Video fingerprint has several properties that make it a perfect tool for fast video matching:

• Lightweight. By design, fingerprints are meant to be processed in massive amounts, therefore, by saving only distinctive features WebKontrol stores much less information on each video.

• Use of visual features. The use of perceptive features means similarity between fingerprints of similar images. Furthermore, it is possible to determine the perceptive similarity of videos only by comparing their fingerprints. This property solves the issue of matching not-exactly-equal videos.

• Indexability. Once encoded, these features may be stored in an index that allows efficient search queries on massive datasets instead of brute-force comparison to every reference video.

Generation of video fingerprints is conducted during a multi-step process that includes the detection and extraction of key video frames and encoding of their visual features to the binary representation. It’s a completely secure process as at no point in time WebKontrol has access to original video content.

WebKontrol video fingerprinting technology is adapted to detect multiple video alterations, distortions, and inconsistencies.

Currently, WebKontrol video fingerprinting technology supports:

• Mirroring
• Dynamic zooming
• Horizontal and vertical sidebars, slight rotation, slight cropping
• Small and large logo, drawings overlays
• Interface elements, ticker, etc.
• Color modifications
• Time inconsistencies such as playback, slight speed increase/decrease
• Video fragments cut
• Deepfake

WebKontrol puts security at the center of the technology design and maintains the highest level of data protection while delivering the services. At no point in time, WebKontrol collects or has access to video files used for the generation of video fingerprints. Video fingerprints are not reversible and cannot be transformed back to original video files. Without metadata, it is also not possible to identify media assets by their respective fingerprints.

With a manual or semi-automated search, it’s almost impossible to identify and verify infringing videos in the limitless universe of videos ever uploaded onto the Internet. Our proprietary video fingerprinting technology solves this problem by enabling precise content identification on the largest resources within seconds. No human can outcompete machine-powered data processing on such a scale.

Digital fingerprinting technology can be applied as a means of video verification to prevent the distribution of unauthorized, duplicate, and non-original content inside existing video ecosystems. Before being published, video content is fingerprinted and matched against a particular video database.