search

LeeJunHyun / Image_Segmentation

Pytorch implementation of U-Net, R2U-Net, Attention U-Net, and Attention R2U-Net.
2.66k stars 594 forks source link

Correction of evaluation metric #59

Closed MELSunny closed 1 year ago

MELSunny commented 4 years ago

tested on pytorch==1.2.0 ISIC 2018 dataset first loop result[All default]: Epoch [1/150], Loss: 570.2478, [Training] Acc: 0.8824, SE: 0.7623, SP: 0.9695, PC: 0.7704, F1: 0.6987, JS: 0.5823, DC: 0.6987 [Validation] Acc: 0.8595, SE: 0.5696, SP: 0.9477, PC: 0.7414, F1: 0.5573, JS: 0.4587, DC: 0.5573 Best R2AttU_Net model score : 1.0160

tested on pytorch==1.5.0 ISIC 2018 dataset first loop result[All default]: Epoch [1/200], Loss: 454.9332, [Training] Acc: 0.9083, SE: 0.7878, SP: 0.9756, PC: 0.8060, F1: 0.7414, JS: 0.6310, DC: 0.7414 [Validation] Acc: 0.8288, SE: 0.3145, SP: 0.9982, PC: 0.6754, F1: 0.3814, JS: 0.2991, DC: 0.3814

MELSunny commented 4 years ago

Correction of evaluation metric tested on pytorch==1.2.0 and pytorch==1.5.0

Hammer-888 commented 2 years ago

tested on pytorch==1.2.0 ISIC 2018 dataset first loop result[All default]: Epoch [1/150], Loss: 570.2478, [Training] Acc: 0.8824, SE: 0.7623, SP: 0.9695, PC: 0.7704, F1: 0.6987, JS: 0.5823, DC: 0.6987 [Validation] Acc: 0.8595, SE: 0.5696, SP: 0.9477, PC: 0.7414, F1: 0.5573, JS: 0.4587, DC: 0.5573 Best R2AttU_Net model score : 1.0160

tested on pytorch==1.5.0 ISIC 2018 dataset first loop result[All default]: Epoch [1/200], Loss: 454.9332, [Training] Acc: 0.9083, SE: 0.7878, SP: 0.9756, PC: 0.8060, F1: 0.7414, JS: 0.6310, DC: 0.7414 [Validation] Acc: 0.8288, SE: 0.3145, SP: 0.9982, PC: 0.6754, F1: 0.3814, JS: 0.2991, DC: 0.3814

but it only works on first epoch in my computer, with torch vision is 1.11.0