TAPe RVS
TAPe RVS — Search Video by Video
TAPe RVS is a Search Video by Video service (Reverse Video Search, RVS) for finding and identifying video fragments in large video archives.
The video itself is used as the search query: users can find the source, a duplicate or modified copy of a video, as well as a specific fragment within an indexed video archive.
Features
- search video by video fragment;
- video identification from a fragment of 2 seconds or longer;
- source identification;
- duplicate and modified-copy detection;
- finding specific scenes and fragments in a video archive;
- comparison of large video archives;
- video archive indexing;
- TV and digital video stream monitoring;
- API and Widget integration.
Video archive search
TAPe RVS is designed for organizations working with large volumes of video data and needing fast search across their own video archives.
Video from proprietary or third-party sources can be indexed for subsequent search. TAPe converts video into a compact data structure designed for search, comparison and analysis.
Approximately 1 hour of video can require about 1 MB in TAPe representation.
Operation within the customer infrastructure
For corporate and sensitive video archives, an architecture can be used in which the original video remains within the customer’s infrastructure.
Video can be indexed locally using a TAPe Converter. A compact TAPe index is transferred to RVS and used for video search and matching.
This approach allows organizations to build Search Video by Video systems without transferring the original video to external infrastructure.
Business use cases
TAPe RVS can be used for:
- finding scenes and materials in large video archives;
- media monitoring;
- TV and digital video stream monitoring;
- source identification and content reuse;
- duplicate and modified-copy detection;
- management of corporate and public video archives;
- video content distribution monitoring;
- building a video search service over a video archive;
- integrating video search into existing IT systems.
Who can use it
- TV networks and news agencies;
- video platforms and streaming services;
- organizations managing large video archives;
- public and corporate archives;
- sports media platforms;
- advertising and marketing companies;
- e-learning platforms;
- organizations requiring automated video search and monitoring.
Technology
TAPe RVS is built on the Theory of Active Perception (TAPe), developed by Comexp.
TAPe analyzes video dynamics and converts video into a compact data structure for search and comparison.
TAPe RVS specifications:
- search for a video fragment in a large video archive — approximately 60 ms;
- approximately 1 MB of TAPe representation per 1 hour of video;
- approximately 10 seconds to convert 1 hour of video;
- identification accuracy — up to 99.9%;
- memory usage — approximately 2 MB RAM.
The approach is robust to transcoding, cropping, noise and reduced video quality.
- Select a TAPe RVS subscription in Yandex Cloud Marketplace.
- Open TAPe RVS and sign in or create an account.
- Connect your own video archive.
- For your own archive, index the video and create a searchable database.
- Search by video fragment through the RVS interface or integrate video search into your application using the API.
TAPe RVS is a SaaS service. No software installation on a virtual machine or local computer is required.
Tariffs
$2.00 / 1 month Prepaid | ||
Mid $10.00 / 1 month Prepaid | $10.00 / 1 month Prepaid | |
Pro $50.00 / 1 month Prepaid | $50.00 / 1 month Prepaid |
- Finding scenes or video fragments in large video archives.
- Finding the source of a video from a short fragment.
- Detecting duplicate and modified copies of videos.
- Monitoring TV and digital video streams.
- Media monitoring and tracking video content distribution.
- Managing corporate and public video archives.
- Building a video search service for users of a website or application.
- Integrating Search Video by Video into existing IT systems through API.
- Working with sensitive video archives without transferring the original video to external infrastructure.