Cats and Dogs

[gjwgit]

Package the keras/tensorflow cats/dog image classification model.

https://www.datasciencecentral.com/profiles/blogs/dogs-vs-cats-image-classification-with-deep-learning-using

https://tensorflow.rstudio.com/blog/keras-image-classification-on-small-datasets.html

[gjwgit]

Bug for py version on dsvm:

$ ml demo image-classification-py
An error was encountered:

Traceback (most recent call last):
  File "demo.py", line 17, in <module>
    loaded_model = load_model(top_model_path)
  File "/anaconda/envs/py35/lib/python3.5/site-packages/keras/models.py", line 237, in load_model
    with h5py.File(filepath, mode='r') as f:
  File "/anaconda/envs/py35/lib/python3.5/site-packages/h5py/_hl/files.py", line 269, in __init__
    fid = make_fid(name, mode, userblock_size, fapl, swmr=swmr)
  File "/anaconda/envs/py35/lib/python3.5/site-packages/h5py/_hl/files.py", line 99, in make_fid
    fid = h5f.open(name, flags, fapl=fapl)
  File "h5py/_objects.pyx", line 54, in h5py._objects.with_phil.wrapper
  File "h5py/_objects.pyx", line 55, in h5py._objects.with_phil.wrapper
  File "h5py/h5f.pyx", line 78, in h5py.h5f.open
OSError: Unable to open file (unable to open file: name = '/home/dlvmadmin/.mlhub/image-classification-py/bottleneck_fc_model.h5', errno = 2, error message = 'No such file or directory', flags = 0, o_flags = 0)

[gjwgit]

Bug for the R version. Probably needs to specify install into the user's ~/R/.... folder for missing packages.

$ ml configure image-classification-r
Configuration will take place using '/home/gjw/.mlhub/image-classification-r/configure.sh'.

An error was encountered:

Installing packages into ‘/usr/local/lib/R/site-library’
(as ‘lib’ is unspecified)
Warning in install.packages(install) :
  'lib = "/usr/local/lib/R/site-library"' is not writable
Error in install.packages(install) : unable to install packages
Execution halted

[gjwgit]

Another R version issue, once configured:

gjw@dsvm01:~$ ml demo image-classification-r
An error was encountered:

Attaching package: ‘dplyr’

The following objects are masked from ‘package:stats’:

    filter, lag

The following objects are masked from ‘package:base’:

    intersect, setdiff, setequal, union

Loading required package: ggplot2
Error in py_call_impl(callable, dots$args, dots$keywords) : 
  TypeError: ('Keyword argument not understood:', 'data_format')

Detailed traceback: 
  File "/anaconda/envs/py35/lib/python3.5/site-packages/keras/models.py", line 243, in load_model
    model = model_from_config(model_config, custom_objects=custom_objects)
  File "/anaconda/envs/py35/lib/python3.5/site-packages/keras/models.py", line 317, in model_from_config
    return layer_module.deserialize(config, custom_objects=custom_objects)
  File "/anaconda/envs/py35/lib/python3.5/site-packages/keras/layers/__init__.py", line 55, in deserialize
    printable_module_name='layer')
  File "/anaconda/envs/py35/lib/python3.5/site-packages/keras/utils/generic_utils.py", line 144, in deserialize_keras_object
    list(custom_objects.items())))
  File "/anaconda/envs/py35/lib/python3.5/site-packages/keras/models.py", line 1349, in from_config
    layer = layer_module.deserialize(conf, custom_objects=custom_objects)
  File "/anaconda/envs/py35/lib/python3.5/site-packages/keras/layers/__init_
Calls: load_model_hdf5 -> do.call -> <Anonymous> -> py_call_impl -> .Call
Execution halted

[ZhouFang928]

I didn't meet the above error message. The results are as shown below:

dlvmadmin@dlvmubuntu03:~$ ml configure image-classification-r Configuration will take place using '/home/dlvmadmin/.mlhub/image-classification-r/configure.sh'. Once configured run the demonstration:

$ ml demo image-classification-r

dlvmadmin@dlvmubuntu03:~$ ml demo image-classification-r

Predict Image Classes

Found 1000 images belonging to 2 classes. Image Predicted Actual Error 1 cat.1501.jpg 0 0
2 cat.1502.jpg 0 0
3 cat.1503.jpg 0 0
4 cat.1504.jpg 0 0
5 cat.1505.jpg 1 0 <---- 6 cat.1506.jpg 0 0
7 cat.1507.jpg 0 0
8 cat.1508.jpg 1 0 <---- 9 cat.1509.jpg 1 0 <---- 10 cat.1510.jpg 1 0 <---- 11 cat.1511.jpg 1 0 <---- 12 cat.1512.jpg 1 0 <---- 13 cat.1513.jpg 1 0 <---- 14 cat.1514.jpg 1 0 <---- 15 cat.1515.jpg 0 0
16 cat.1516.jpg 0 0
17 cat.1517.jpg 1 0 <---- 18 cat.1518.jpg 1 0 <---- 19 cat.1519.jpg 0 0
20 cat.1520.jpg 0 0

Model Loss and Accuracy

$loss [1] 0.2575455

$acc [1] 0.89

Confusion Matrix

Confusion Matrix and Statistics

      Reference

Prediction 0 1 0 230 252 1 270 248

           Accuracy : 0.478           
             95% CI : (0.4466, 0.5095)
No Information Rate : 0.5             
P-Value [Acc > NIR] : 0.9227          

              Kappa : -0.044          

Mcnemar's Test P-Value : 0.4568

        Sensitivity : 0.4960          
        Specificity : 0.4600   

[ZhouFang928]

The results for image-classification-py are as shown below:

dlvmadmin@dlvmubuntu03:~$ ml demo image-classification-py Found 1000 images belonging to 2 classes.

Predict image classes

       Image  Predicted  Actual

0 cat.1947.jpg 0 0 1 cat.1990.jpg 0 0 2 cat.1642.jpg 0 0 3 cat.1999.jpg 0 0 4 cat.1513.jpg 0 0 5 cat.1508.jpg 1 0 6 cat.1540.jpg 1 0 7 cat.1673.jpg 0 0 8 cat.1638.jpg 0 0 9 cat.1913.jpg 0 0 10 cat.1908.jpg 0 0 11 cat.1664.jpg 0 0 12 cat.1570.jpg 0 0 13 cat.1720.jpg 0 0 14 cat.1746.jpg 0 0 15 cat.1560.jpg 0 0 16 cat.1979.jpg 0 0 17 cat.1937.jpg 0 0 18 cat.1639.jpg 0 0 19 cat.1976.jpg 0 0

Accuracy

50/50 [==============================] - 2s 44ms/step acc: 87.00%

Confusion Matrix

[[412 88] [ 42 458]] dlvmadmin@dlvmubuntu03:~$

[ZhouFang928]

python tutorial https://blog.keras.io/building-powerful-image-classification-models-using-very-little-data.html