I'm looking computer vision developer who has a experience in field of foreground-background separation.
Currently I have a solution based on deeplabv3.
I'm going to upgrade current model, so want background removal model with more accuracy.
The model is TFlite for consuming into android platform, so the model size should be less than 10M.
The model should be segment with the following object list.
hat and cap
If you have experience in this field, please contact me.