3DFaceMorph

The repository provides sample code for the paper "3D Face Morphing Attacks: Generation, Vulnerability and Detection", accepted in IEEE TBIOM and available at Arxiv (https://arxiv.org/abs/2201.03454). The code sample can be executed by running completeScript.sh on Ubuntu 18.04 with python3.6 and pymeshlab, cv2, mmcv, facenet_pytorch, and DeepFace packages. The code proceeds in the following order.

1. The code sample first generates color and depth maps from point clouds of two individuals.
2. It then generates a color image and depth map of the morph.
3. The point cloud is generated for the face morphing image and depth.(AnkurShukla005_Raghu005_translatedScaled.ply)
4. A small manual cleanup step of the generated face morphing point cloud (AnkurShukla005_Raghu005_translatedScaledCleaned_ascii.ply)
5. A hole-filled point cloud is generated, which could be cleaned further (AnkurShukla005_Raghu005_translatedScaledCleaned_ascii_averaged.ply)