This is a django app which has a pretrained image classification model.The users can upload images of women dresses and can get which class does this dress belongs. we can build a recommendation system by extending the mcdel based on the user data. The dataset used for this model is fashion dataset which have all the images stored in s3. we will create our dataset by running the script. please check wiki for installation and training the model.The accuracy of the model with train data is 73.7% and 72 with validation data.
check the detailed explanation in https://github.com/mohanpillimitla/hello/wiki
The jupyternotebook in which the model is trained https://github.com/mohanpillimitla/hello/blob/master/uploadimage/model.ipynb
you can train the model again by opening the above model.ipynb in google colab and tune the hyperparameters