you359/Keras-FasterRCNN

It doesn't train,Epoch 1/2000,and stop for one hour

HYG-777 opened this issue · 2 comments

python train_frcnn.py -p VOCdevkit2007 --network vgg --num_epochs 20
Using TensorFlow backend.
Parsing annotation files
Processing 000531.xml: 100%|██████████████| 6798/6798 [00:01<00:00, 3668.26it/s]
Training images per class:
{'bg': 0, 'car': 32750}
Num classes (including bg) = 2
Config has been written to config.pickle, and can be loaded when testing to ensure correct results
Num train samples 6119
Num test samples 611
loading weights from vgg16_weights_tf_dim_ordering_tf_kernels.h5
Starting training
Epoch 1/20

Although I have downloaded the pre-trained model, this is still the case,but when i use resnet,it trains
Processing 000531.xml: 100%|██████████████| 6798/6798 [00:02<00:00, 2971.03it/s]
Training images per class:
{'bg': 0, 'car': 32750}
Num classes (including bg) = 2
Config has been written to config.pickle, and can be loaded when testing to ensure correct results
Num train samples 6119
Num test samples 611
loading weights from resnet50_weights_tf_dim_ordering_tf_kernels.h5
Could not load pretrained model weights. Weights can be found in the keras application folder https://github.com/fchollet/keras/tree/master/keras/applications
Starting training
Epoch 1/2000
1/1000 [..............................] - ETA: 7:27:19 - rpn_cls: 1.9539 - rp 2/1000 [..............................] - ETA: 5:25:22 - rpn_cls: 3.4903 - rp 3/1000 [..............................] - ETA: 5:03:22 - rpn_cls: 3.7469 - rp 4/1000 [..............................] - ETA: 4:26:33 - rpn_cls: 3.7732 - rp 5/1000 [..............................] - ETA: 4:10:56 - rpn_cls: 3.9350 - rp 6/1000 [..............................] - ETA: 3:56:59 - rpn_cls: 3.9521 - rp 7/1000 [..............................] - ETA: 3:48:50 - rpn_cls: 3.9520 - rp 8/1000 [..............................] - ETA: 3:40:04 - rpn_cls: 3.9512 - rp 9/1000 [..............................] - ETA: 3:36:31 - rpn_cls: 3.9798 - rp 10/1000 [..............................] - ETA: 3:34:02 - rpn_c

so ,what happends