Thoroughly eliminate photo hacking of facial recognition! "Impersonation detection technology" to determine whether it is biometric or not.
The "impostor detection technology" utilizing "joint patent technology" with Panasonic is a biometric detection (biometric sensing) security element technology that can be used for "fraudulent impersonation and hacking countermeasures" such as "facial recognition" and "eKYC (online identity verification)." It can be performed using "visible light cameras" such as web cameras (USB cameras) and built-in PC cameras, achieving high-precision judgments through deep learning. It prevents "fraudulent impersonation" through "fake face detection" that can be determined with just one frame, and "action detection" that detects specified actions such as "blinking" and "head orientation" using time-series data. We have compressed the learning model, achieving a processing speed of about 100ms per frame even with CPU processing (on our Windows machine with an 8th generation Core i7 CPU). We offer both an SDK and an application, with the application supporting CSV output and recording functions as standard, making it capable of handling not only real-time verification. The application is available in both "camera input version" and "file input version."
Inquire About This Product
basic information
◆ "Impersonation Detection Technology" Four Features 1. **[Fake Detection] Detects smartphone impersonation** Using deep learning, it detects fake face hacking by analyzing photos or face images projected on smartphone screens. 2. **[Action Detection] Prevents impersonation by requiring users to perform random actions** Based on time-series data from camera input, it instructs users to perform specific actions and can determine "impersonation detection" by detecting whether the prescribed actions are performed. 3. **Can be implemented on any camera, including smartphone cameras!** This "impersonation detection technology" is less dependent on specific cameras, allowing for easy implementation of "impersonation detection" using many widely available cameras, such as those built into smartphones or USB webcams. 4. **Optimal for edge AI, capable of fast processing with only CPU using a deep learning model** By compressing the deep learning training model as much as possible, it enables fast processing even without a GPU. With recent Core i5/i7 processors, processing can be achieved at around 100ms per frame. (Processing speed may vary depending on the type of device.)
Price information
Please contact us. For more details, please visit our website.
Delivery Time
Applications/Examples of results
Usage: ◆ Impersonation judgment during facial recognition ◆ Photo capture during eKYC Others
News about this product（1）
Company information
We are researching and developing element technologies related to human sensing using image processing techniques based on deep learning, and providing them. ◆ Impersonation Detection Technology This technology combines fake detection, which determines whether a face in a photo or on a smartphone is real or fake, with action detection, which assesses whether a user is performing the instructed action or impersonating someone. https://bio-check.pas-ta.io/ ◆ Eye Detection/Blink Detection Technology This highly versatile technology allows for easy measurement of "blinks," "eye feature points," and "viewing distance (the distance from the camera to the eye)" using a web camera or PC device camera. https://pupil.pas-ta.io/ ◆ Gaze Detection Technology This technology analyzes camera input or video files to detect a person's "gaze direction" and "face orientation." https://eyetrack.pas-ta.io/ ◆ Iris Authentication Technology This authentication technology uses the unique pattern of the "iris" in a person's eye, which does not overlap with others. Currently, it is only provided to partners involved in the joint development of iris capture cameras. https://iris.pas-ta.io/