Flask Image Classification

This repo comprises of training DenseNet from scratch on cifar10 and then using Flask for Model serving.

The directory is defined as

For Training

Run Training Notebook file

For Inference

we need Cifar-10 data for inference. Execute the below to generate 10 random images from cifar-10 test data.

cd data
python generate_valid_data.py

Run Inference Notebook file

Model Serving on Flask.

For Serving this model as a website. First generate test images by executing above commands and then.

python flask_api/app.py