A Binary image classifier using keras that classifies images of ship and truck
Notebook used for training the model can be found on the repo
The model is deployed as an API on heroku.

Link to the API https://truckshipclassification.herokuapp.com/detect

Send a POST request with the image with name "image" as form-data

Response from the API

Trained on images of ships and trucks from cifar-10 dataset.
The dataset can be found on the repo
Dataset is trained on both Convolutional Neural Network and model using Transfer learning on VGG16

Used three block VGG model architechture, achieved 80% accuracy on the test dataset

Line plots of loss and accuracy Learning curves for the model

VGG16 model has 2 parts, the feature extraction part of the model that is made up of VGG blocks, and classifier part of the model made using fully connected layers and the output layer
Used the feature extraction part of the model and added a new classifier part.
The weights of all the convolutional layers are fixed during training, and only trained the fully connected layers to learn to interpret the features extracted from the model for binary classification
The model achieved very impressive results with a classification accuracy of 97% on the test dataset

Line plots of loss and accuracy Learning curves for the model

The model is deployed on heroku using the flask web framework.

• Flask
• Keras
• Numpy

Please feel free to raise any issue...