Train on my custom dataset with no labels

mhamilton723 / STEGO

Unsupervised Semantic Segmentation by Distilling Feature Correspondences

MIT License

711 stars 142 forks source link

Train on my custom dataset with no labels #56

Closed HalvesChen closed 1 year ago

HalvesChen commented 1 year ago

Hi! Thank you for your amazing work! I want to run the train_segmentation.py code, but I don't have labels. What do I need to change in the validation part(such as validation_step(), validation_epoch_end()) of the trainer?

mhamilton723 commented 1 year ago

Thanks for reaching out! Please follow the instructions here and omit the label directory

https://github.com/mhamilton723/STEGO#bringing-your-own-data