Open MagicXiaoJing opened 4 years ago
sorry to trouble you again. but i really cannot get the similar result even from src code from this git. I encounter the overfitting problem and the mask show is nothing like yous in paper. the dataset i use is oulu
this is some info log from mine:
APCER BPCER ACER 52.5601 52.5601 52.5601
2020-07-04 14:28:00,863-INFO: Epoch: 3, Iter: 24200, ETA: 58.13 hours, lr: 0.000500, Best_score: 94.88 | loss_cls: 0.00479, loss_reg: 0.00562, loss_tir: 0.00000, loss: 0.01041 2020-07-04 14:28:29,737-INFO: Epoch: 3, Iter: 24210, ETA: 822.08 hours, lr: 0.000500, Best_score: 94.88 | loss_cls: 0.00310, loss_reg: 0.00552, loss_tir: 0.00000, loss: 0.00862'
this is my mask and orig img :
really appreciate for your answer , thanks
sorry to trouble you again. but i really cannot get the similar result even from src code from this git. I encounter the overfitting problem and the mask show is nothing like yous in paper. the dataset i use is oulu
this is some info log from mine:
APCER BPCER ACER 52.5601 52.5601 52.5601
2020-07-04 14:28:00,863-INFO: Epoch: 3, Iter: 24200, ETA: 58.13 hours, lr: 0.000500, Best_score: 94.88 | loss_cls: 0.00479, loss_reg: 0.00562, loss_tir: 0.00000, loss: 0.01041 2020-07-04 14:28:29,737-INFO: Epoch: 3, Iter: 24210, ETA: 822.08 hours, lr: 0.000500, Best_score: 94.88 | loss_cls: 0.00310, loss_reg: 0.00552, loss_tir: 0.00000, loss: 0.00862'
this is my mask and orig img :
really appreciate for your answer , thanks