Missing shape in roi_pool (layer_input to fc6)

[sean-wong]

The roi_pool (input to fc) seems to be missing its shape Tensor("pool_5:0", dtype=float32) and causing the training to break. Any ideas (I installed tensorflow 1.0). Thanks!

Traceback (most recent call last): File "./tools/train_net.py", line 91, in network = get_network(args.network_name) File "Faster-RCNN_TF/tools/../lib/networks/factory.py", line 30, in get_network return networks.VGGnet_train() File "Faster-RCNN_TF/tools/../lib/networks/VGGnet_train.py", line 20, in init self.setup() File "Faster-RCNN_TF/tools/../lib/networks/VGGnet_train.py", line 83, in setup .fc(4096, name='fc6') File "Faster-RCNN_TF/tools/../lib/networks/network.py", line 25, in layer_decorated layer_output = op(self, layer_input, *args, **kwargs) File "Faster-RCNN_TF/tools/../lib/networks/network.py", line 243, in fc feed_in, dim = (input, int(input_shape[-1])) TypeError: int returned non-int (type NoneType)

[osimeoni]

Did you find out what is causing the error ?

[forwardwfg]

i meet the same problem now. Did you solve it ?