The objective of this work is to remove mask objects in facial images. We break the problem into two stages: mask object detection and image completion of the removed mask region.
- For this task we used Mask RCNN to detect the object (facemask) and generate binary segmentation maps and implemented this model using detectron2
-
The goal of this model is to remove the mask and complete the left behind region in a way that is both structural and appearance wise consistent with the ground truth image.
-
For this task, We are going through some state-of-the-art architectures such as GAN (generative adversarial network), autoencoders etc.
👷👷👷👷
*We are currently working on this to find the best architecture for this task (facial inpainting)*
- Mask Object Detection and Segmentation
- Image Completion
- Model Deployment
- For Pretrained models of object detection and segmentation Detectron2
- For Datal Labeling labelflow
- To Create Simulated Masked Dataset MaskTheFace
- The Reference Research paper