I need to be able to upload an image, to select part of the image and input it into another photo all on a mobile phone.
So far I can train MobilenetV2 on my own dataset and for it to reconginse my images. I can't get instance segmentation to work either with the coco dataset or with my own. I have tried with Mask RCNN, but no joy. Looking at trying SOLO next.
Need someone who can help make this a reality with the ability to update images and training of the datasets as new images become needed.
Ability to select an item in image.
For item to be removed and inserted into another image.
To be able to train the dataset to increase accuracy.