/udacity-ai-flower-image-classification

Transfer learning for classifying flower images

Primary LanguageHTMLMIT LicenseMIT

Flower classifier project

Overview

The Flower classifier project is part of the AI Programming with Python Nanodegree program. It consists on applying Transfer learning over a known architecture (such as VGG or Densenet) in order to classify flower images.

This repository consists on a two part project:

  • Jupyter notebook showcasing the training of a neural network in order to classify flower images.
  • Transfer of the jupyter experiment code to a set of python scripts that will let us train a neural network and predict given an image, what kind of flower category it is.

Dependencies

To install these dependencies with pip, you can issue pip install -r requirements.txt.

Part 1: Jupyter notebook

You can find the Jupyter notebook experiment on the file ImageClassifierProject.ipynb.

Part 2: Python scripts usage

Training

Example of use:

python train.py --data_dir './flowers' --arch='vgg16' --save_dir './' --learning_rate 0.001 --hidden_units 5

00 --gpu --epochs 10
Training model using GPU
Transfer learning process starting
- Loading dataset
- Creating transfer learning model
- Training model
Epoch: 1/10 ---  Training Loss: 4.5241   Validation Loss: 4.4263   Validation Accuracy: 0.0920  
Epoch: 1/10 ---  Training Loss: 4.3233   Validation Loss: 3.9992   Validation Accuracy: 0.1994  
Epoch: 1/10 ---  Training Loss: 3.9725   Validation Loss: 3.6344   Validation Accuracy: 0.2477  
Epoch: 1/10 ---  Training Loss: 3.4935   Validation Loss: 3.0354   Validation Accuracy: 0.3591  
Epoch: 1/10 ---  Training Loss: 3.3410   Validation Loss: 2.7064   Validation Accuracy: 0.4233  
Epoch: 1/10 ---  Training Loss: 2.9503   Validation Loss: 2.5111   Validation Accuracy: 0.4116  
Epoch: 1/10 ---  Training Loss: 2.7216   Validation Loss: 2.1833   Validation Accuracy: 0.4674  
Epoch: 1/10 ---  Training Loss: 2.6785   Validation Loss: 2.0220   Validation Accuracy: 0.4925  
Epoch: 1/10 ---  Training Loss: 2.4526   Validation Loss: 1.8436   Validation Accuracy: 0.5375  
Epoch: 1/10 ---  Training Loss: 2.3576   Validation Loss: 1.7647   Validation Accuracy: 0.5211  
Epoch: 1/10 ---  Training Loss: 2.2228   Validation Loss: 1.4945   Validation Accuracy: 0.5947  
Epoch: 1/10 ---  Training Loss: 2.1641   Validation Loss: 1.4633   Validation Accuracy: 0.5872  
Epoch: 1/10 ---  Training Loss: 1.9855   Validation Loss: 1.3383   Validation Accuracy: 0.6408  
Epoch: 1/10 ---  Training Loss: 1.7683   Validation Loss: 1.3295   Validation Accuracy: 0.6500  
Epoch: 1/10 ---  Training Loss: 1.7327   Validation Loss: 1.2229   Validation Accuracy: 0.6605  
Epoch: 1/10 ---  Training Loss: 1.6465   Validation Loss: 1.2392   Validation Accuracy: 0.6616  
Epoch: 1/10 ---  Training Loss: 1.6383   Validation Loss: 1.2337   Validation Accuracy: 0.6804  
Epoch: 1/10 ---  Training Loss: 1.6286   Validation Loss: 1.1750   Validation Accuracy: 0.6512  
Epoch: 1/10 ---  Training Loss: 1.5494   Validation Loss: 1.1669   Validation Accuracy: 0.6603  
Epoch: 1/10 ---  Training Loss: 1.5168   Validation Loss: 1.0019   Validation Accuracy: 0.7090  
Epoch: 2/10 ---  Training Loss: 0.8161   Validation Loss: 1.0385   Validation Accuracy: 0.7206  
Epoch: 2/10 ---  Training Loss: 1.5290   Validation Loss: 0.9711   Validation Accuracy: 0.7405  
Epoch: 2/10 ---  Training Loss: 1.3511   Validation Loss: 0.9697   Validation Accuracy: 0.7288  
Epoch: 2/10 ---  Training Loss: 1.3157   Validation Loss: 0.8521   Validation Accuracy: 0.7553  
Epoch: 2/10 ---  Training Loss: 1.2058   Validation Loss: 0.9171   Validation Accuracy: 0.7366  
Epoch: 2/10 ---  Training Loss: 1.3185   Validation Loss: 0.9952   Validation Accuracy: 0.7234  
Epoch: 2/10 ---  Training Loss: 1.3003   Validation Loss: 0.9049   Validation Accuracy: 0.7393  
Epoch: 2/10 ---  Training Loss: 1.4582   Validation Loss: 0.9711   Validation Accuracy: 0.7330  
Epoch: 2/10 ---  Training Loss: 1.4177   Validation Loss: 0.9168   Validation Accuracy: 0.7423  
Epoch: 2/10 ---  Training Loss: 1.2598   Validation Loss: 0.8650   Validation Accuracy: 0.7571  
Epoch: 2/10 ---  Training Loss: 1.2975   Validation Loss: 0.8628   Validation Accuracy: 0.7720  
Epoch: 2/10 ---  Training Loss: 1.2762   Validation Loss: 0.8819   Validation Accuracy: 0.7547  
Epoch: 2/10 ---  Training Loss: 1.2740   Validation Loss: 0.8902   Validation Accuracy: 0.7646  
Epoch: 2/10 ---  Training Loss: 1.2818   Validation Loss: 0.7304   Validation Accuracy: 0.7958  
Epoch: 2/10 ---  Training Loss: 1.3364   Validation Loss: 0.7714   Validation Accuracy: 0.7833  
Epoch: 2/10 ---  Training Loss: 1.4652   Validation Loss: 0.7397   Validation Accuracy: 0.7913  
Epoch: 2/10 ---  Training Loss: 1.2195   Validation Loss: 0.7325   Validation Accuracy: 0.7953  
Epoch: 2/10 ---  Training Loss: 1.2516   Validation Loss: 0.7320   Validation Accuracy: 0.7914  
Epoch: 2/10 ---  Training Loss: 1.0585   Validation Loss: 0.6811   Validation Accuracy: 0.8208  
Epoch: 2/10 ---  Training Loss: 1.3765   Validation Loss: 0.6549   Validation Accuracy: 0.8094  
Epoch: 2/10 ---  Training Loss: 1.1998   Validation Loss: 0.7030   Validation Accuracy: 0.8015  
Epoch: 3/10 ---  Training Loss: 1.0401   Validation Loss: 0.7760   Validation Accuracy: 0.7790  
Epoch: 3/10 ---  Training Loss: 1.0453   Validation Loss: 0.6509   Validation Accuracy: 0.8184  
Epoch: 3/10 ---  Training Loss: 1.0721   Validation Loss: 0.7419   Validation Accuracy: 0.8073  
Epoch: 3/10 ---  Training Loss: 1.2632   Validation Loss: 0.7977   Validation Accuracy: 0.7780  
Epoch: 3/10 ---  Training Loss: 1.1062   Validation Loss: 0.7366   Validation Accuracy: 0.8064  
Epoch: 3/10 ---  Training Loss: 1.0814   Validation Loss: 0.6938   Validation Accuracy: 0.8169  
Epoch: 3/10 ---  Training Loss: 1.0440   Validation Loss: 0.6200   Validation Accuracy: 0.8289  
Epoch: 3/10 ---  Training Loss: 0.9234   Validation Loss: 0.6789   Validation Accuracy: 0.8142  
Epoch: 3/10 ---  Training Loss: 1.1727   Validation Loss: 0.6208   Validation Accuracy: 0.8190  
Epoch: 3/10 ---  Training Loss: 1.0728   Validation Loss: 0.5579   Validation Accuracy: 0.8496  
Epoch: 3/10 ---  Training Loss: 1.0369   Validation Loss: 0.6790   Validation Accuracy: 0.8076  
Epoch: 3/10 ---  Training Loss: 1.2384   Validation Loss: 0.6214   Validation Accuracy: 0.8232  
Epoch: 3/10 ---  Training Loss: 1.0957   Validation Loss: 0.6241   Validation Accuracy: 0.8116  
Epoch: 3/10 ---  Training Loss: 1.0939   Validation Loss: 0.6694   Validation Accuracy: 0.8078  
Epoch: 3/10 ---  Training Loss: 1.1558   Validation Loss: 0.6248   Validation Accuracy: 0.8261  
Epoch: 3/10 ---  Training Loss: 1.0271   Validation Loss: 0.5708   Validation Accuracy: 0.8333  
Epoch: 3/10 ---  Training Loss: 0.9878   Validation Loss: 0.7196   Validation Accuracy: 0.8037  
Epoch: 3/10 ---  Training Loss: 1.0556   Validation Loss: 0.6368   Validation Accuracy: 0.8198  
Epoch: 3/10 ---  Training Loss: 1.0548   Validation Loss: 0.6252   Validation Accuracy: 0.8395  
Epoch: 3/10 ---  Training Loss: 1.2056   Validation Loss: 0.5994   Validation Accuracy: 0.8161  
Epoch: 4/10 ---  Training Loss: 0.4902   Validation Loss: 0.6776   Validation Accuracy: 0.8181  
Epoch: 4/10 ---  Training Loss: 1.0293   Validation Loss: 0.6899   Validation Accuracy: 0.7926  
Epoch: 4/10 ---  Training Loss: 0.9641   Validation Loss: 0.6532   Validation Accuracy: 0.8321  
Epoch: 4/10 ---  Training Loss: 0.9909   Validation Loss: 0.5931   Validation Accuracy: 0.8416  
Epoch: 4/10 ---  Training Loss: 0.9592   Validation Loss: 0.6485   Validation Accuracy: 0.8244  
Epoch: 4/10 ---  Training Loss: 0.8746   Validation Loss: 0.6175   Validation Accuracy: 0.8307  
Epoch: 4/10 ---  Training Loss: 0.8555   Validation Loss: 0.5619   Validation Accuracy: 0.8373  
Epoch: 4/10 ---  Training Loss: 0.9293   Validation Loss: 0.5682   Validation Accuracy: 0.8371  
Epoch: 4/10 ---  Training Loss: 0.7355   Validation Loss: 0.6248   Validation Accuracy: 0.8280  
Epoch: 4/10 ---  Training Loss: 1.0139   Validation Loss: 0.6130   Validation Accuracy: 0.8467  
Epoch: 4/10 ---  Training Loss: 1.1700   Validation Loss: 0.6166   Validation Accuracy: 0.8383  
Epoch: 4/10 ---  Training Loss: 1.1018   Validation Loss: 0.5491   Validation Accuracy: 0.8467  
Epoch: 4/10 ---  Training Loss: 1.0548   Validation Loss: 0.5358   Validation Accuracy: 0.8539  
Epoch: 4/10 ---  Training Loss: 0.8418   Validation Loss: 0.6189   Validation Accuracy: 0.8094  
Epoch: 4/10 ---  Training Loss: 0.9617   Validation Loss: 0.6105   Validation Accuracy: 0.8424  
Epoch: 4/10 ---  Training Loss: 1.0914   Validation Loss: 0.5859   Validation Accuracy: 0.8400  
Epoch: 4/10 ---  Training Loss: 0.9141   Validation Loss: 0.5382   Validation Accuracy: 0.8539  
Epoch: 4/10 ---  Training Loss: 1.0310   Validation Loss: 0.5688   Validation Accuracy: 0.8355  
Epoch: 4/10 ---  Training Loss: 0.8192   Validation Loss: 0.5153   Validation Accuracy: 0.8560  
Epoch: 4/10 ---  Training Loss: 0.9503   Validation Loss: 0.5405   Validation Accuracy: 0.8424  
Epoch: 4/10 ---  Training Loss: 0.8330   Validation Loss: 0.5133   Validation Accuracy: 0.8514  
Epoch: 5/10 ---  Training Loss: 0.8424   Validation Loss: 0.5432   Validation Accuracy: 0.8397  
Epoch: 5/10 ---  Training Loss: 0.9096   Validation Loss: 0.4973   Validation Accuracy: 0.8665  
Epoch: 5/10 ---  Training Loss: 0.8689   Validation Loss: 0.5464   Validation Accuracy: 0.8460  
Epoch: 5/10 ---  Training Loss: 0.9716   Validation Loss: 0.5554   Validation Accuracy: 0.8500  
Epoch: 5/10 ---  Training Loss: 0.8961   Validation Loss: 0.5781   Validation Accuracy: 0.8446  
Epoch: 5/10 ---  Training Loss: 0.8859   Validation Loss: 0.5305   Validation Accuracy: 0.8403  
Epoch: 5/10 ---  Training Loss: 0.8808   Validation Loss: 0.5342   Validation Accuracy: 0.8482  
Epoch: 5/10 ---  Training Loss: 0.9323   Validation Loss: 0.5474   Validation Accuracy: 0.8431  
Epoch: 5/10 ---  Training Loss: 0.9082   Validation Loss: 0.5128   Validation Accuracy: 0.8566  
Epoch: 5/10 ---  Training Loss: 0.9470   Validation Loss: 0.6160   Validation Accuracy: 0.8292  
Epoch: 5/10 ---  Training Loss: 0.9218   Validation Loss: 0.5506   Validation Accuracy: 0.8587  
Epoch: 5/10 ---  Training Loss: 0.9220   Validation Loss: 0.4864   Validation Accuracy: 0.8693  
Epoch: 5/10 ---  Training Loss: 0.9793   Validation Loss: 0.5425   Validation Accuracy: 0.8446  
Epoch: 5/10 ---  Training Loss: 0.8019   Validation Loss: 0.5670   Validation Accuracy: 0.8487  
Epoch: 5/10 ---  Training Loss: 0.8844   Validation Loss: 0.5253   Validation Accuracy: 0.8587  
Epoch: 5/10 ---  Training Loss: 0.8455   Validation Loss: 0.5534   Validation Accuracy: 0.8484  
Epoch: 5/10 ---  Training Loss: 0.7557   Validation Loss: 0.5506   Validation Accuracy: 0.8547  
Epoch: 5/10 ---  Training Loss: 0.9100   Validation Loss: 0.5440   Validation Accuracy: 0.8409  
Epoch: 5/10 ---  Training Loss: 0.6570   Validation Loss: 0.5884   Validation Accuracy: 0.8436  
Epoch: 5/10 ---  Training Loss: 1.0641   Validation Loss: 0.5728   Validation Accuracy: 0.8523  
Epoch: 6/10 ---  Training Loss: 0.5182   Validation Loss: 0.5264   Validation Accuracy: 0.8578  
Epoch: 6/10 ---  Training Loss: 0.8283   Validation Loss: 0.5531   Validation Accuracy: 0.8381  
Epoch: 6/10 ---  Training Loss: 0.8770   Validation Loss: 0.5282   Validation Accuracy: 0.8695  
Epoch: 6/10 ---  Training Loss: 0.9311   Validation Loss: 0.5624   Validation Accuracy: 0.8431  
Epoch: 6/10 ---  Training Loss: 0.8559   Validation Loss: 0.5656   Validation Accuracy: 0.8349  
Epoch: 6/10 ---  Training Loss: 0.8030   Validation Loss: 0.5138   Validation Accuracy: 0.8524  
Epoch: 6/10 ---  Training Loss: 0.7748   Validation Loss: 0.4734   Validation Accuracy: 0.8698  
Epoch: 6/10 ---  Training Loss: 0.7913   Validation Loss: 0.5209   Validation Accuracy: 0.8623  
Epoch: 6/10 ---  Training Loss: 0.8849   Validation Loss: 0.5456   Validation Accuracy: 0.8599  
Epoch: 6/10 ---  Training Loss: 0.7675   Validation Loss: 0.5601   Validation Accuracy: 0.8506  
Epoch: 6/10 ---  Training Loss: 0.7468   Validation Loss: 0.5556   Validation Accuracy: 0.8532  
Epoch: 6/10 ---  Training Loss: 0.7957   Validation Loss: 0.5196   Validation Accuracy: 0.8611  
Epoch: 6/10 ---  Training Loss: 1.0252   Validation Loss: 0.5189   Validation Accuracy: 0.8599  
Epoch: 6/10 ---  Training Loss: 0.6669   Validation Loss: 0.4954   Validation Accuracy: 0.8705  
Epoch: 6/10 ---  Training Loss: 0.9000   Validation Loss: 0.5114   Validation Accuracy: 0.8607  
Epoch: 6/10 ---  Training Loss: 0.8523   Validation Loss: 0.5739   Validation Accuracy: 0.8421  
Epoch: 6/10 ---  Training Loss: 0.9316   Validation Loss: 0.5077   Validation Accuracy: 0.8614  
Epoch: 6/10 ---  Training Loss: 1.0028   Validation Loss: 0.5202   Validation Accuracy: 0.8599  
Epoch: 6/10 ---  Training Loss: 0.7707   Validation Loss: 0.5458   Validation Accuracy: 0.8515  
Epoch: 6/10 ---  Training Loss: 0.8097   Validation Loss: 0.5067   Validation Accuracy: 0.8650  
Epoch: 6/10 ---  Training Loss: 0.8044   Validation Loss: 0.5274   Validation Accuracy: 0.8535  
Epoch: 7/10 ---  Training Loss: 0.8557   Validation Loss: 0.5971   Validation Accuracy: 0.8340  
Epoch: 7/10 ---  Training Loss: 0.7239   Validation Loss: 0.5632   Validation Accuracy: 0.8446  
Epoch: 7/10 ---  Training Loss: 0.8529   Validation Loss: 0.5249   Validation Accuracy: 0.8734  
Epoch: 7/10 ---  Training Loss: 0.7584   Validation Loss: 0.5262   Validation Accuracy: 0.8608  
Epoch: 7/10 ---  Training Loss: 0.7297   Validation Loss: 0.5012   Validation Accuracy: 0.8674  
Epoch: 7/10 ---  Training Loss: 0.7272   Validation Loss: 0.5116   Validation Accuracy: 0.8690  
Epoch: 7/10 ---  Training Loss: 0.8906   Validation Loss: 0.4883   Validation Accuracy: 0.8578  
Epoch: 7/10 ---  Training Loss: 0.7464   Validation Loss: 0.4799   Validation Accuracy: 0.8666  
Epoch: 7/10 ---  Training Loss: 0.7479   Validation Loss: 0.4872   Validation Accuracy: 0.8690  
Epoch: 7/10 ---  Training Loss: 0.8648   Validation Loss: 0.5515   Validation Accuracy: 0.8515  
Epoch: 7/10 ---  Training Loss: 0.8160   Validation Loss: 0.5568   Validation Accuracy: 0.8503  
Epoch: 7/10 ---  Training Loss: 0.7839   Validation Loss: 0.5054   Validation Accuracy: 0.8666  
Epoch: 7/10 ---  Training Loss: 0.7023   Validation Loss: 0.4961   Validation Accuracy: 0.8743  
Epoch: 7/10 ---  Training Loss: 0.8118   Validation Loss: 0.4819   Validation Accuracy: 0.8786  
Epoch: 7/10 ---  Training Loss: 0.7241   Validation Loss: 0.5642   Validation Accuracy: 0.8416  
Epoch: 7/10 ---  Training Loss: 0.7860   Validation Loss: 0.5041   Validation Accuracy: 0.8689  
Epoch: 7/10 ---  Training Loss: 0.9155   Validation Loss: 0.5481   Validation Accuracy: 0.8669  
Epoch: 7/10 ---  Training Loss: 0.8139   Validation Loss: 0.5315   Validation Accuracy: 0.8590  
Epoch: 7/10 ---  Training Loss: 0.8745   Validation Loss: 0.4958   Validation Accuracy: 0.8719  
Epoch: 7/10 ---  Training Loss: 0.9250   Validation Loss: 0.4885   Validation Accuracy: 0.8753  
Epoch: 8/10 ---  Training Loss: 0.3667   Validation Loss: 0.4923   Validation Accuracy: 0.8734  
Epoch: 8/10 ---  Training Loss: 0.7574   Validation Loss: 0.4654   Validation Accuracy: 0.8777  
Epoch: 8/10 ---  Training Loss: 0.7410   Validation Loss: 0.5014   Validation Accuracy: 0.8619  
Epoch: 8/10 ---  Training Loss: 0.9187   Validation Loss: 0.5242   Validation Accuracy: 0.8659  
Epoch: 8/10 ---  Training Loss: 0.8629   Validation Loss: 0.4856   Validation Accuracy: 0.8695  
Epoch: 8/10 ---  Training Loss: 0.6717   Validation Loss: 0.4625   Validation Accuracy: 0.8758  
Epoch: 8/10 ---  Training Loss: 0.7567   Validation Loss: 0.4898   Validation Accuracy: 0.8722  
Epoch: 8/10 ---  Training Loss: 0.6260   Validation Loss: 0.5089   Validation Accuracy: 0.8703  
Epoch: 8/10 ---  Training Loss: 0.8813   Validation Loss: 0.5204   Validation Accuracy: 0.8698  
Epoch: 8/10 ---  Training Loss: 0.5997   Validation Loss: 0.5648   Validation Accuracy: 0.8550  
Epoch: 8/10 ---  Training Loss: 0.7693   Validation Loss: 0.5256   Validation Accuracy: 0.8689  
Epoch: 8/10 ---  Training Loss: 0.6395   Validation Loss: 0.4813   Validation Accuracy: 0.8758  
Epoch: 8/10 ---  Training Loss: 0.7131   Validation Loss: 0.4861   Validation Accuracy: 0.8866  
Epoch: 8/10 ---  Training Loss: 0.8395   Validation Loss: 0.4636   Validation Accuracy: 0.8876  
Epoch: 8/10 ---  Training Loss: 0.6777   Validation Loss: 0.4341   Validation Accuracy: 0.8846  
Epoch: 8/10 ---  Training Loss: 0.7886   Validation Loss: 0.4834   Validation Accuracy: 0.8749  
Epoch: 8/10 ---  Training Loss: 0.7331   Validation Loss: 0.4404   Validation Accuracy: 0.8821  
Epoch: 8/10 ---  Training Loss: 0.9220   Validation Loss: 0.4273   Validation Accuracy: 0.8839  
Epoch: 8/10 ---  Training Loss: 0.6955   Validation Loss: 0.4922   Validation Accuracy: 0.8686  
Epoch: 8/10 ---  Training Loss: 0.7880   Validation Loss: 0.5042   Validation Accuracy: 0.8693  
Epoch: 8/10 ---  Training Loss: 0.8415   Validation Loss: 0.4983   Validation Accuracy: 0.8746  
Epoch: 9/10 ---  Training Loss: 0.6811   Validation Loss: 0.4510   Validation Accuracy: 0.8731  
Epoch: 9/10 ---  Training Loss: 0.8056   Validation Loss: 0.4749   Validation Accuracy: 0.8741  
Epoch: 9/10 ---  Training Loss: 0.6411   Validation Loss: 0.5176   Validation Accuracy: 0.8689  
Epoch: 9/10 ---  Training Loss: 0.7912   Validation Loss: 0.4725   Validation Accuracy: 0.8789  
Epoch: 9/10 ---  Training Loss: 0.6924   Validation Loss: 0.5230   Validation Accuracy: 0.8666  
Epoch: 9/10 ---  Training Loss: 0.7140   Validation Loss: 0.5146   Validation Accuracy: 0.8689  
Epoch: 9/10 ---  Training Loss: 0.7340   Validation Loss: 0.4991   Validation Accuracy: 0.8662  
Epoch: 9/10 ---  Training Loss: 0.6187   Validation Loss: 0.5200   Validation Accuracy: 0.8678  
Epoch: 9/10 ---  Training Loss: 0.8513   Validation Loss: 0.5240   Validation Accuracy: 0.8574  
Epoch: 9/10 ---  Training Loss: 0.6897   Validation Loss: 0.5444   Validation Accuracy: 0.8604  
Epoch: 9/10 ---  Training Loss: 0.7727   Validation Loss: 0.4819   Validation Accuracy: 0.8810  
Epoch: 9/10 ---  Training Loss: 0.6696   Validation Loss: 0.4501   Validation Accuracy: 0.8798  
Epoch: 9/10 ---  Training Loss: 0.8389   Validation Loss: 0.4305   Validation Accuracy: 0.8897  
Epoch: 9/10 ---  Training Loss: 0.8320   Validation Loss: 0.4557   Validation Accuracy: 0.8731  
Epoch: 9/10 ---  Training Loss: 0.8385   Validation Loss: 0.4881   Validation Accuracy: 0.8698  
Epoch: 9/10 ---  Training Loss: 0.7892   Validation Loss: 0.4375   Validation Accuracy: 0.8936  
Epoch: 9/10 ---  Training Loss: 0.9552   Validation Loss: 0.4515   Validation Accuracy: 0.8851  
Epoch: 9/10 ---  Training Loss: 0.6781   Validation Loss: 0.4859   Validation Accuracy: 0.8802  
Epoch: 9/10 ---  Training Loss: 0.8650   Validation Loss: 0.4482   Validation Accuracy: 0.8900  
Epoch: 9/10 ---  Training Loss: 0.7905   Validation Loss: 0.5032   Validation Accuracy: 0.8717  
Epoch: 10/10 ---  Training Loss: 0.3353   Validation Loss: 0.4511   Validation Accuracy: 0.8854  
Epoch: 10/10 ---  Training Loss: 0.7632   Validation Loss: 0.4671   Validation Accuracy: 0.8827  
Epoch: 10/10 ---  Training Loss: 0.6682   Validation Loss: 0.4933   Validation Accuracy: 0.8753  
Epoch: 10/10 ---  Training Loss: 0.6472   Validation Loss: 0.4498   Validation Accuracy: 0.8921  
Epoch: 10/10 ---  Training Loss: 0.5817   Validation Loss: 0.4337   Validation Accuracy: 0.8825  
Epoch: 10/10 ---  Training Loss: 0.6271   Validation Loss: 0.4777   Validation Accuracy: 0.8683  
Epoch: 10/10 ---  Training Loss: 0.8009   Validation Loss: 0.4882   Validation Accuracy: 0.8730  
Epoch: 10/10 ---  Training Loss: 0.6388   Validation Loss: 0.4790   Validation Accuracy: 0.8717  
Epoch: 10/10 ---  Training Loss: 0.6678   Validation Loss: 0.4899   Validation Accuracy: 0.8713  
Epoch: 10/10 ---  Training Loss: 0.6793   Validation Loss: 0.4432   Validation Accuracy: 0.8888  
Epoch: 10/10 ---  Training Loss: 0.8604   Validation Loss: 0.4551   Validation Accuracy: 0.8722  
Epoch: 10/10 ---  Training Loss: 0.9162   Validation Loss: 0.5010   Validation Accuracy: 0.8698  
Epoch: 10/10 ---  Training Loss: 0.8016   Validation Loss: 0.5503   Validation Accuracy: 0.8650  
Epoch: 10/10 ---  Training Loss: 0.8277   Validation Loss: 0.4950   Validation Accuracy: 0.8725  
Epoch: 10/10 ---  Training Loss: 0.6964   Validation Loss: 0.4745   Validation Accuracy: 0.8739  
Epoch: 10/10 ---  Training Loss: 0.8663   Validation Loss: 0.4734   Validation Accuracy: 0.8777  
Epoch: 10/10 ---  Training Loss: 0.6483   Validation Loss: 0.4806   Validation Accuracy: 0.8773  
Epoch: 10/10 ---  Training Loss: 0.7241   Validation Loss: 0.5104   Validation Accuracy: 0.8777  
Epoch: 10/10 ---  Training Loss: 0.7043   Validation Loss: 0.4704   Validation Accuracy: 0.8753  
Epoch: 10/10 ---  Training Loss: 0.7900   Validation Loss: 0.5235   Validation Accuracy: 0.8649  
Epoch: 10/10 ---  Training Loss: 0.6053   Validation Loss: 0.4816   Validation Accuracy: 0.8729  
- Saving model into disk
- Model saved at ./checkpoint.pth
Training process completed.

Prediction

python predict.py --checkpoint "./checkpoint.pth" --input flowers/train/1/image_06735.jpg

Prediction results
==================================
Class: pink primrose, Probability: 0.9863
Class: pelargonium, Probability: 0.0098
Class: tree mallow, Probability: 0.0021
Class: hibiscus, Probability: 0.0010
Class: californian poppy, Probability: 0.0003