carly666
commented
6 years ago The same problem here. It's confusing because I once trained successfully. But now, the loss_cls and loss_bbox is always 0.
Open 673096604 opened 6 years ago
carly666
commented
6 years ago The same problem here. It's confusing because I once trained successfully. But now, the loss_cls and loss_bbox is always 0.
iter: 10 / 80000, total loss: 0.8165, rpn_loss_cls: 0.4789, rpn_loss_box: 0.3374, loss_cls: 0.0002, loss_box: 0.0000, lr: 0.001000 speed: 0.446s / iter iter: 20 / 80000, total loss: 0.3478, rpn_loss_cls: 0.2360, rpn_loss_box: 0.1119, loss_cls: 0.0000, loss_box: 0.0000, lr: 0.001000 speed: 0.325s / iter iter: 30 / 80000, total loss: 0.6299, rpn_loss_cls: 0.2044, rpn_loss_box: 0.4255, loss_cls: 0.0000, loss_box: 0.0000, lr: 0.001000 speed: 0.284s / iter iter: 40 / 80000, total loss: 0.8028, rpn_loss_cls: 0.1882, rpn_loss_box: 0.6146, loss_cls: 0.0000, loss_box: 0.0000, lr: 0.001000 speed: 0.262s / iter iter: 50 / 80000, total loss: 0.5187, rpn_loss_cls: 0.1347, rpn_loss_box: 0.3841, loss_cls: 0.0000, loss_box: 0.0000, lr: 0.001000 speed: 0.249s / iter iter: 60 / 80000, total loss: 0.3959, rpn_loss_cls: 0.1608, rpn_loss_box: 0.2351, loss_cls: 0.0000, loss_box: 0.0000, lr: 0.001000 speed: 0.240s / iter iter: 70 / 80000, total loss: 0.2651, rpn_loss_cls: 0.1698, rpn_loss_box: 0.0954, loss_cls: 0.0000, loss_box: 0.0000, lr: 0.001000 speed: 0.235s / iter iter: 80 / 80000, total loss: 0.1934, rpn_loss_cls: 0.1188, rpn_loss_box: 0.0746, loss_cls: 0.0000, loss_box: 0.0000, lr: 0.001000 speed: 0.231s / iter iter: 90 / 80000, total loss: 1.5419, rpn_loss_cls: 0.1451, rpn_loss_box: 1.3968, loss_cls: 0.0000, loss_box: 0.0000, lr: 0.001000 speed: 0.227s / iter iter: 100 / 80000, total loss: 0.4818, rpn_loss_cls: 0.1473, rpn_loss_box: 0.3345, loss_cls: 0.0000, loss_box: 0.0000, lr: 0.001000 speed: 0.224s / iter iter: 110 / 80000, total loss: 0.9081, rpn_loss_cls: 0.1285, rpn_loss_box: 0.7796, loss_cls: 0.0000, loss_box: 0.0000, lr: 0.001000 speed: 0.222s / iter iter: 120 / 80000, total loss: 0.3060, rpn_loss_cls: 0.1813, rpn_loss_box: 0.1247, loss_cls: 0.0000, loss_box: 0.0000, lr: 0.001000 speed: 0.220s / iter iter: 130 / 80000, total loss: 0.1560, rpn_loss_cls: 0.0727, rpn_loss_box: 0.0833, loss_cls: 0.0000, loss_box: 0.0000, lr: 0.001000 speed: 0.218s / iter iter: 140 / 80000, total loss: 0.2424, rpn_loss_cls: 0.1368, rpn_loss_box: 0.1056, loss_cls: 0.0000, loss_box: 0.0000, lr: 0.001000 speed: 0.216s / iter iter: 150 / 80000, total loss: 0.3982, rpn_loss_cls: 0.1361, rpn_loss_box: 0.2621, loss_cls: 0.0000, loss_box: 0.0000, lr: 0.001000
why loss_box:.0.0000 ,loss_box:0.0000 ? my picture size 1280*720 ,Are they related?