JagadishSivakumar / Image-Classifier-ml5.js

Image Classifier using ml5.js

Image-Classifier-ml5.js

• ML5.js is a pre-trained library in JS for image classifictaion https://ml5js.org/

• MobileNet is an efective convolutional Neural Network for Mobile Apps , https://arxiv.org/abs/1704.04861

• The ML5.js library gets all the dataset from : http://image-net.org/ ImageNet is a repository of around 15 million images gathered from google searches, Wikipedia, and other open sources.

Working of ML5.js

ML5.js is a pretrained image classifier that runs in browser It is an opensourced classfier developer by ML Developers, It works on the basis of Supervised Learning Supervised Learning - Supervised learning is the Data mining task of inferring a function from labeled training data set.

• The Data set needs an input as an image
• The data set contains Input, image in (png, jpg) and its label in a .csv file

The Files

index.html file , the main initialization file

Examples and Setup : https://ml5js.org/docs/getting-started.html

sketch.js , the main code for placing the image as input for classifier and steps of prediction

Running in Browser

To run the image classifier in browser , open the index.html file in local-host To enable prediction, replace the image in puffin = createImg('path', imageReady);
here puffin is a variable Test Images of Penguin, Puffin is included To Test more images add images to the images folder . Assign the user input to the variable value to make prediction instant