Smart Extraction: Targeted Deepfake Detection | identifAI
Smart Extraction: Targeted Deepfake Detection for Faces, Video and Audio
With Smart Extraction, you have greater control over which video frames and audio segments identifAI analyzes for deepfakes. From prioritizing frames containing faces to selecting specific moments in video and audio, you can tailor content selection to your verification needs.
In short: Smart Extraction is a new identifAI feature, available now, with three capabilities: Specific Frames containing Faces (video), Specific Moments (video) and Specific Moments (audio).
Why targeted extraction matters
Video and audio files contain different types of information, and automatic sampling may not select the moments most relevant to a particular verification task. Smart Extraction lets you focus on the frames or segments that matter, tailoring content selection to your needs.
1. identifAI New Video Smart Extraction Feature - Faces
When analyzing face-related manipulation, frames without visible faces may offer limited information for face classification.
Smart Extraction builds on evenly spaced frame sampling by checking whether each selected frame contains a face. If no face is detected, the system searches a surrounding time interval for an alternative frame, keeping samples as evenly distributed across the video as possible.
Best for: videos in which people appear intermittently, and workflows that involve identifying and classifying visible subjects.
2. identifAI New Video Smart Extraction Feature - Specific Moments
Sometimes you already know which part of a video needs closer examination. Smart Extraction lets you select specific frames or segments instead of relying entirely on automatic selection, so reviewers can focus on the relevant scenes, interactions or moments.
Best for: investigators, media reviewers and teams examining specific moments in a recording.
3. identifAI New Audio Smart Extraction Feature - Specific Moments
You can also select specific segments of an audio file. This is useful when a particular sentence, voice passage or section of a recording raises concerns.
Best for: teams investigating suspicious calls, voice messages, interviews or other audio recordings.
How Smart Extraction works
- Upload: paste a media link or upload a file from your browser.
- Configure: choose your analysis options and extraction mode: select key frames with faces or select specific moments.
- Confirm: confirm you selection and that your submission respects the guidelines tp start the analysis.
- Review: receive the final verdict (artificial score).
Who can benefit?
Smart Extraction supports media verification workflows across security, fraud prevention, trust and safety, journalism and digital investigations. It gives teams more control over the content they submit for analysis. As with any detection tool, results depend on the content submitted.
A word from our CTO
With Smart Extraction, you determine which faces, video moments and audio segments identifAI analyzes. This enhances your control over deepfake analysis, letting you focus on the parts of a file that matter most to you. We are constantly improving our technology to give you a more comprehensive and efficient way to verify the truth behind digital media.
Marco Prati, CTO, identifAI
Try Smart Extraction with identifAI
To see how targeted video and audio analysis fits your verification workflow, contact sales@identifai.net or visit identifai.net.
Frequently asked questions
What is Smart Extraction?
Smart Extraction is an identifAI feature that lets you control which video frames or audio segments are selected for deepfake analysis.
What are its three capabilities?
Smart Extraction: Faces (video), Smart Extraction: Specific Moments (video) and Smart Extraction: Specific Moments (audio).
How does face extraction work?
It checks that each sampled frame contains a face. If not, it looks for an alternative frame nearby, keeping samples evenly spread across the video.
Can I choose which part of a video or audio file to analyze?
Yes. Specific Moments lets you select the frames, segments or audio passages to analyze instead of relying on automatic selection.
Does Smart Extraction work on audio?
Yes, through Specific Moments. Face extraction applies to video only.
Does Smart Extraction change how the verdict is produced?
No. It gives you control over which content is analyzed. The verdict is still returned as the platform’s artificial score.
Is Smart Extraction available now?
Yes. Contact sales@identifai.net or visit identifai.net to learn more.
talk to a human expert
Tell us about your business. We'll come back to you within one business day.
No sales pitch. Just a conversation.
We stand for truth
Gartner does not endorse any vendor, product or service depicted in its researchpublications, and does not advise technology users to select only those vendors with thehighest ratings or other designation. Gartner research publications consist of the opinions ofGartner’s research organization and should not be construed as statements of fact. Gartnerdisclaims all warranties, expressed or implied, with respect to this research, including anywarranties of merchantability or fitness for a particular purpose. GARTNER is a registeredtrademark and service mark of Gartner, Inc. and/or its affiliates in the U.S. and internationallyand is used herein with permission. All rights reserved.