flower-classifier-webapp's Introduction

FLOWER-CLASSIFIER-WEBAPP

This is an neural network webapp visualizing the training of the network and testing accuracy of 98.6% accuracy. The neural network uses pretrained resnet152 and then trained to classify images of flowers. It is built using Pytorch framework using Python as primary language. The webapp is built using flask.

Dataset used :

102 Category Flower Dataset

http://www.robots.ox.ac.uk/~vgg/data/flowers/102/index.html
Maria-Elena Nilsback and Andrew Zisserman

Neural Network used :

Resnet152
Network Layers in Pytorch

You can dowload ResNet152 here
You can download overall modified model here

Refresher on Neural Network :

Gradient Descent
Backpropogation

Flow :

Input image is fed and transformed using : commons.py
Inference is done by : inference.py

Complete flow :

Run on windows -

Make sure you have installed Python , Pytorch and flask.

First download all the folders and files
git clone https://github.com/souravs17031999/FLOWER-CLASSIFIER-WEBAPP.git
Download pretrained weights and keep it in the same Project directory Download here.
Then open the command prompt (or powershell) and change the directory to the path where all the files are located.
cd FLOWER-CLASSIFIER-WEBAPP
Now run the following commands -

set FLASK_APP=flower.py

flask run

This will firstly download the models and then start the local web server.

now go to the local server something like this - http://127.0.0.1:5000/ and see the result and explore.

@creator - sourav kumar

Recommend Projects