Explain the forensic watermarking workflow at the level of OTT platform from insertion to detection

The over-the-top (OTT) content space has seen a significant boom in recent years. With growing demand for streaming platforms and content, video piracy has also become a major concern for content creators and operators. While distribution of DRM protected content has  become absolutely essential, forensic watermarking techniques are often used to provide security against piracy and leakage.

In this technique, imperceptible user-specific information is embedded in the digital media which can be detected from the pirated copy to identify the source of infringement. The forensic watermarking solution follows three basic steps as given below:

1. Preparation: In the first step, the video asset undergoes an encoding process wherein a watermarked frame is added to the original uncompressed content. Some watermarking solutions can carry out encoding and watermark embedding simultaneously.
2. Embedding: A watermark embedder then adds imperceptible identifiers or owner/user information to the encoded content. This information can only be encrypted using a secret key which is unique to every copy of encoded content. In the session-based video watermarking process, which is typically used by OTT providers, session information, such as user ID, content ID, device information, etc., is converted into binary format to get an output that combines DASH and HLS segments, thereby creating a unique watermark for each playback session. Based on the requirements of the OTT platform or content owner, the embedding method can vary from using an embedder library, watermark packing service, or CLI embedder. Multiple embedding options are also provided in some solutions.
3. Capture and extraction: Once a watermark is embedded, the content can be distributed to the intended audience. In case of leakage, the pirated content is captured to gather information about the infringement. A watermark detector analyzes each frame of the pirated content to identify the original watermark. The same key which is used during embedding is used for the detection process. If the detector finds the watermark payload, the session information can be used by the OTT platform to identify the exact source of leakage. Based on this information, the platform or content creator can take several anti-piracy measures and also initiate legal proceedings against the source of piracy.

OTT platforms tend to use a cloud-based vendor which bundles multi-DRM service and video watermarking service in a single workflow to not only add the two security measures together but also to find the user who causes the leak in the shortest possible time.