Classify pictures with Deep Learning Model
This study is to learn to use, create and train deep learning CNN models (ResNet / VGG).
An association which helps and hosts a lot of dogs need to classify pictures of their animals. To do that, it is possible to use CNN models by training on Stanford Dogs Dataset (http://vision.stanford.edu/aditya86/ImageNetDogs/).
These is my works about this study :
01_OC_IML_Project_6_explorations.ipynb : exploration
02_OC_IML_Project_6_my_neural_network.ipynb & 03_OC_IML_Project_6_my_neural_network_2.ipynb : Creation / test of a neural network CNN from scratch.
04_OC_IML_Project_6_10breeds.ipynb & 05_OC_IML_Project_6_3breeds.ipynb : VGG-16 Neural Network transfert learning
06_OC_IML_Project_6_3breeds_ResNet50.ipynb & 07_OC_IML_Project_6_10breeds_ResNet50.ipynb : ResNet-50 transfert learning
08_OC_IML_Project_6_my_neural_network_DAugm_01.ipynb : data augmentation on my neural network from scratch
09_OC_IML_Project_6_10breeds_ResNet50_script.ipynb : notebook to create and test script
- script_predict_breed.py : command line script to predict breed (ResNet50 10 breeds)
- ResNet50_TL_10b_9.h5 : transfert learning model
- df_dogs.pkl : contains list of breeds (and other stuff about training pictures)