resnet-50 with ohem error

[whmin]

when i run ./experiments/scripts/faster_rcnn_end2end.sh 0 ResNet-50 pascal_voc,i got a error:

i did not change the original code,but only replace the "num_classes" and "num_output", is there anyone can help me?Thank you very much!!! I am waiting here!

[liuliu66]

@whmin it seems that some indexes contain non-integers. you can check the error location in code and add int() to these indexes.
And where do you get the ResNet50 pre-trained model to implement it? I also want to try ResNet50 for my task. Thank you!

[quibbler123]

this error caused by numpy's version. try this modify in ./lib/roi_data_layer/layer.py

• for x in [cls_prob[i, int(label)] for i, label in enumerate(labels)]]

[maozezhong]

@liuliu66 have you got the RetNet50 pre-trained model for faster-rcnn-ohem? hope for your reply~