how to retrain my own cocodatasets based on modal_final.pth pretrained???

facebookresearch / maskrcnn-benchmark

Fast, modular reference implementation of Instance Segmentation and Object Detection algorithms in PyTorch.

MIT License

9.3k stars 2.49k forks source link

how to retrain my own cocodatasets based on modal_final.pth pretrained??? #987

Open tangze5258 opened 5 years ago

tangze5258 commented 5 years ago

how to retrain my own cocodatasets based on modal_final.pth pretrained??? i add some new cocodatasets ,have modified WEIGHT: "./model_final.pth(pretrained)", and then redo the "python train_net.py --config-file *.yaml",but it do not train, total training time =0,just skip the train process. . how to deal with it ? thx..

th512 commented 5 years ago

hi, you need to delete or modify this file with the name of last_checkpoint.

tangze5258 commented 5 years ago

hi, you need to delete or modify this file with the name of last_checkpoint.

@th512 thank you for your advice.Actually, i delete optimizer， scheduler and iteration dict and it did work when run train_net.py, but finally, has no new model_final.pth file generated,what's wrong ? thx....

th512 commented 5 years ago

@tangze5258 what are your running commands? you just need to delete last_checkpoint. "~/.torch/models/*.pkl",Is there a model in this path?