-
In this repository, hand segmentation model is created using pre-trained weights.
-
For this project, segmentation_models(https://github.com/qubvel/segmentation_models) package is used to create U-net with efficientnetb3 backbone and then applied pre-trained weights.
-
I also provided Keras .H5 model with U-net architecture which takes input as (320,320) image and gives Segmentaed image with same height and width.
-
Here is one example of Hand-Segmenation using this model,
Google Drive : Hand-Segmentation Model