Single Image Liveness
This page describes the single image liveness feature of Liveness.
A Single Image Liveness (or 2D Liveness) Detection is a passive, image-based, monocular liveness check which uses only one RBG image as the input media. The Single Image Liveness in Liveness detects the largest face in an uploaded image and returns the results of the spoof score.
Input media requirements
For a Single Image Liveness detection, there should be one image included in the request body. The image file should meet the following standards.
| Item | Minimum requirement | Recommended |
|---|---|---|
| File size | Under 4MB (approximately) | 200KB |
| Minimum Pixels of the face | 32×32 | 200×200 |
| File format | JPG (JPEG), PNG, GIF (The first frame only) | JPG (JPEG), PNG |
| Number of faces in the portrait | More than 1 | 1 only |
The detection logic is defined to proceed with only one face in the image. But images with multiple faces are acceptable to improve the user experience.
Models
There is currently one model available for Single Image Liveness. Any model improvements will not change its name or the recommended threshold value to address the user's API compatibility. Feel comfortable using the same model name and the recommended threshold value.
| Model name | Input media | Recommended threshold |
|---|---|---|
| JCV_2D_I25000 | images: [0]RGB | 0.5 |
Updated about 1 year ago