Image requirements¶
To run the processing the IDLive Face requires an image to match the set of conditions. This conditions might vary depending of pipelines that would be used: PAD or DFD.
DFD-specific requirements are listed in DFD requirements section.
Image format¶
IDLive Face supports JPEG, PNG and BMP formats.
Image compression¶
IDLive Face suppport image compressed up to the JPEG-70. However, while the resulting compression artifacts may be invisible to the naked eye, they may still be significant to the anti-spoofing algorithm. The larger the JPEG number (JPEG 80, JPEG 90, etc.), the less compression performed and therefore the higher quality retained by the image. We recommend minimal compression, preferably of a lossless variety such as PNG.
Image resolution¶
IDLive Face processes the area around the face, so the optimal resolution will depend on the image content. Assuming a portrait orientation, with the face centered and taking at least 1/4 of image area, the minimum image resolution is SVGA (600x800). In general images captured in higher resolution will result in better accuracy.
Image content¶
-
There should only be one main face on the image. It should be fully visible within a frame and fully open without any occlusions. No crop is allowed. Small faces in the background are not taken into account.
-
The minimum size of a face box that can be processed is 150x150 pixels.
-
The padding between the face box and the image's borders should be at least 15 pixels.
-
The distance between the pupils on the face should be at least 50 pixels.
-
The out-of-plane rotation angle (face pitch and yaw) should be no more than ±30 degrees.
Images that do not match these conditions will be rejected. Other conditions can affect the accuracy:
-
Motion blur effect can significantly increase BPCER (the rate of errors classifying a genuine person as a spoof).
-
Texture filtering can significantly increase APCER (the rate of errors allowing impostors through).
-
Spotlights on the faces and nearest surroundings can significantly increase BPCER.
-
Ill-lighted environments and colored light can significantly increase BPCER.
-
Fish-eye lenses are not supported.
-
Sunglasses may be confusing.
DFD requirements¶
Image format¶
-
For DFD usage of lossless or near-lossless compression (PNG or JPEG-100) is highly recommended for best accuracy.
-
Compression ratio below JPEG-70 will result in significant accuracy degradation.
Image resolution¶
- Recommended image resolution is Full HD, also known as FHD or 1080p (1920x1080)
Image content¶
-
The recommended size of a face box is 400x400 pixels. Smaller face boxes can significantly increase BPCER.
-
IDLive Face Deepfake Detection (DFD) pipelines provide protection from digital manipulations, thus filters or any other traces of digital manipulations are not allowed as they will be rejected.
Sample images¶
This section contains image samples separated into 3 categories:
-
Good samples - the images with optimal face size and positioning, lighting conditions.
-
Correct, but not good enough samples - the images which are acceptable but were taken in unoptimal conditions.
-
Incorrect samples - the images which are not acceptable for processing.
Good samples¶
Correct, but not good enough samples¶
A distance between the camera and the face is too big:
Face is partially obscured by headwear:
Incorrect samples¶
Poor lighting conditions:
Headwear obscures face too much:
Wrong head rotation:
Face is too close to the camera:
Face is positioned incorrectly, partially out of frame: