How to fine-tune this model for a semantic segmentation task for human parsing ?

Gaoyiminggithub / Graphonomy

Graphonomy: Universal Human Parsing via Graph Transfer Learning

MIT License

295 stars 68 forks source link

How to fine-tune this model for a semantic segmentation task for human parsing ? #20

Closed rishab-sharma closed 3 years ago

rishab-sharma commented 5 years ago

Can this model be mapped to a single node graph for human parsing ( Semantic Segmentation - 2 classes - human / Background), if yes, how should I calculate the adj matrix for the same. Any help or guidance on the same would be appreciated.

Ahanio commented 4 years ago

Did you figure it out?

Gaoyiminggithub commented 3 years ago

You can set the label to 2 classes - human/ Background, and then train the model. However, our method may not show the superiority with 2 classes only.