Deepdetect Docker Training Error

[seyfullahuysal]

I get the error in the picture when applying the classification example given in the link via docker jupyterlab. https://www.deepdetect.com/platform/docs/introduction/

I placed .caffemodel file and train data folder in that directory “/home/jovyan/models/cats_dogs”

How can i fix this problem?

Another Question When i apply this example https://www.deepdetect.com/server/docs/train-image-classifier/ i received that error

{"status":{"code":400,"msg":"BadRequest","dd_code":1006,"dd_msg":"Service Bad Request Error: failed to locate model template /home/jovyan/models/cats_dogs/SE-ResNet-50/.prototxt"}

I get error via used that steps

mkdir /home/jovyan/models mkdir /home/jovyan/models/cats_dogs

and below files are stored in the “/home/jovyan/models/cats_dogs” directory SE-ResNet-50.caffemodel SE-ResNet-50.prototxt Extracted data from https://www.deepdetect.com/dd/datasets/cats_dogs.zip

docker run -d -p 8080:8080 -v /home/jovyan/models/cats_dogs jolibrain/deepdetect_cpu

curl -X PUT "http://localhost:8080/services/catsdogs" -d '{ "mllib":"caffe", "description":"image classification service", "type":"supervised", "parameters":{ "input":{ "connector":"image", "width":224, "height":224 }, "mllib":{ "template":"SE-ResNet-50", "nclasses":2, "finetuning":true } }, "model":{ "templates":"/home/jovyan/models/cats_dogs", "repository":"/home/jovyan/models/cats_dogs", "weight": "SE-ResNet-50.caffemodel" } }'

What’s mean “template” argument and how can fix?

[alx]

Hi @seyfullahuysal

Thanks for your feedback.

/home/jovyan/models/cats_dogs is not visible from inside your docker volumes.

You'd need to move your model files to $DD_PLATFORM/models/cats_dogs - $DD_PLATFORM is $HOME/deepdetect if you've followed the install instructions on deepdetect.com - and use the following path in your jupyter notebook: /opt/platform/models/cats_dogs

Feel free to come back to us if it's still not working or if you have more questions

[seyfullahuysal]

Hi @alx Thans for your interest

First I need to explain, before ask this question when I run this example I received "no module named dd_widgets" error then dd_widgets installed by me as described here using docker terminal and i rid that error.

Note : Now i dont get "no module named dd_widgets" error in JupyterLab but when run this code

import sys sys.path.append('/home/jovyan/dd_wigets/build/lib/dd_widgets') from dd_widgets import Classification, CSV, Text, Segmentation, Detection, OCR, TSNE_CSV

OR THIS

import sys sys.path.append('/opt/platform/notebooks/ddmodels/train/widgets/') from dd_widgets import Classification, CSV, Text, Segmentation, Detection, OCR, TSNE_CSV

in python what placed to jupyterlab terminal i getting that error

After apply the fixes that you describe i got the same error

- My last situation

- My files in $DD_PLATFORM/models/cats_dogs directory

These are my questions

1. How to define train.prototxt or solver.prototxt files in the following block, how the system detects these files because we only give path as a model_repo ?

   

2. Is training_repo parameter contains just train files or contains tarin and test folders ?

3. What is the value of template parameter to train custom caffe model ?

Thanks for aid