Keras-retinanet: after setup, train the voc2007 data, but can't download the ResNet-50-model.keras.h5

Created on 5 Mar 2018 · 15Comments · Source: fizyr/keras-retinanet

ResNet-50-model.keras.h5
https://github.com/fizyr/keras-models/releases/download/v0.0.1/ResNet-50-model.keras.h5
Downloading data from https://github.com/fizyr/keras-models/releases/download/v0.0.1/ResNet-50-model.keras.h5
Traceback (most recent call last):
File "train.py", line 367, in
main()
File "train.py", line 340, in main
weights = download_imagenet(args.backbone)
File "../../keras_retinanet/models/resnet.py", line 52, in download_imagenet
md5_hash=checksum
File "/usr/lib64/python2.7/site-packages/keras/utils/data_utils.py", line 224, in get_file
raise Exception(error_msg.format(origin, e.errno, e.reason))
Exception: URL fetch failure on https://github.com/fizyr/keras-models/releases/download/v0.0.1/ResNet-50-model.keras.h5: None -- [Errno 110] Connection timed out

how should i do with this ?

Source

JaosonMa

Most helpful comment

model should be moved to ~/.keras/models, good luck

xieenze on 5 Mar 2018

👍2

All 15 comments

i can download the ResNet-50-model.keras.h5 on my windows compter, but i don.t know where should i put the file in? if it works right, where the file will be download to ?
我不知道这个文件下载到哪里？我现在可以已经有这个文件了，但是我不知道Linux下自动下载的时候这个文件会被放在那里？

JaosonMa on 5 Mar 2018

You can use the argument --weights to specify where the h5 file is.

panda9095 on 5 Mar 2018

👍2

you mean this? @panda9095

JaosonMa on 5 Mar 2018

if you are in china , you need to set vpn open and paste the url to brower to download the model.

xieenze on 5 Mar 2018

model should be moved to ~/.keras/models, good luck

xieenze on 5 Mar 2018

👍2

you mean this? @panda9095

You should not modify the default value of the flag. You should pass the desired value on the command line.

de-vri-es on 5 Mar 2018

@de-vri-es

JaosonMa on 5 Mar 2018

@de-vri-es i try this way , but at last this error

Creating model, this may take a second...
/home/jansonm/keras-retinanet/data/keras_models/ResNet-50-model.keras.h5
Traceback (most recent call last):
  File "train.py", line 368, in <module>
    main()
  File "train.py", line 344, in main
    model, training_model, prediction_model = create_models(backbone_retinanet=retinanet, backbone=args.backbone, num_classes=train_generator.num_classes(), weights=weights, multi_gpu=args.multi_gpu)
  File "train.py", line 76, in create_models
    model            = model_with_weights(backbone_retinanet(num_classes, backbone=backbone, nms=True), weights=weights, skip_mismatch=True)
  File "train.py", line 55, in model_with_weights
    model.load_weights(weights, by_name=True, skip_mismatch=skip_mismatch)
  File "/usr/lib64/python2.7/site-packages/keras/engine/topology.py", line 2642, in load_weights
    raise ImportError('`load_weights` requires h5py.')
ImportError: `load_weights` requires h5py

JaosonMa on 5 Mar 2018