search

chenxinpeng / SSD_scene_text_detection

Detect text in natural images with SSD, Single Shot Detection
224 stars 103 forks source link

Error when loading the deploy.prototxt #5

Closed ghost closed 7 years ago

ghost commented 7 years ago

Hi, I just tried to setup the system using a recent version of caffe from github. When I execute

#init caffee
model_def = curr_dir+'/training_file/deploy.prototxt'
model_weights = curr_dir+'/training_file/VGG_scenetext_SSD_300x300_iter_60000.caffemodel'

net = caffe.Net(model_def,      # defines the structure of the model
                model_weights,  # contains the trained weights
                caffe.TEST)     # use test mode (e.g., don't perform dropout)

I receive this error:

Net('/home/eugen/projects/deep_learning/SSD_scene-text-detection/training_file/deploy.prototxt', 1, weights='/home/eugen/projects/deep_learning/SSD_scene-text-detection/training_file/VGG_scenetext_SSD_300x300_iter_60000.caffemodel')

[libprotobuf ERROR google/protobuf/text_format.cc:245] Error parsing text-format caffe.NetParameter: 758:14: Message type "caffe.LayerParameter" has no field named "norm_param".

F0105 00:19:47.013073 21762 upgrade_proto.cpp:88] Check failed: ReadProtoFromTextFile(param_file, param) Failed to parse NetParameter file: /home/eugen/projects/deep_learning/SSD_scene-text-detection/training_file/deploy.prototxt

Do I have to use the SSD-Caffe fork? WIll try it now and see if it helps...

ghost commented 7 years ago

yes, the reason was, hat one has to use the SSD-form of caffe: https://github.com/weiliu89/caffe/tree/ssd